Study and development of Ardupilot missions for fixed-wings

Study and development of Ardupilot missions for fixed-wings Document: Report Author: Marc Falomir Vil`a Director: Oriol Lordan Gonzalez Bachelor Degree: Grau en Enginyeria en Vehicles Aeroespacials Semester: Spring/Prorogation, 2023. T´ıtol del TFE Abstract Although Unmanned Aerial Vehicles (UAV) have been developed since the early 20th century, it is only over the last two decades that they have gained prominence in the civilian market. This has opened the doors to a whole new world of possibilities where the UAV s are offering a cheaper, faster and most effective solution. In this project, the first prototype of a fixed wing autonomous UAV is developed to act as a relaying node between a ground station and an other UAV , which is operating Beyond Visual Line Of Sight ( BVLOS ); thus ensuring connectivity between the two ends of the chain without the use of satellite communications or other complex ground infrastructure. The resulting UAV system is composed of two main parts: the airframe and the avionics. Three sub-systems can also be distinguished in the avionics group, being those the propulsion plant, the autopilot and the communications system; whose common feature is to be 100% Open Source. A functional prototype of the described UAV is built, and both the design parameters of the vehicle and configuration of the autopilot is validated through flight tests. All the design decisions, configuration parameters and building steps are documented in this report. The developed prototype aims demonstrate the viability of an aerial system built out of affordable materials, simple building techniques, and low-cost accessible open source hardware and software; that provides a solution for the rapid and efficient deployment of an airborne system capable of operating BVLOS. I T´ıtol del TFE Contents 1 Introduction 1 1.1 Object ................................................. 1 1.2 Scope ................................................. 1 1.3 Requirements ............................................. 2 1.4 Justification .............................................. 2 2 State of the art 5 2.1 Market overview ........................................... 5 2.2 Battery-powered UAVs ........................................ 6 2.3 Application insights ......................................... 6 2.3.1 Aerial photography and videography ............................ 7 2.3.2 Precision agriculture ..................................... 7 2.3.3 Surveying, mapping and inspection ............................. 7 2.3.4 Delivery and logistics .................................... 8 2.4 Classification and regulation of Unmanned Aerial Vehicles (UAVs) ............... 8 2.4.1 Drone regulations in Spain ................................. 9 2.5 Current autonomous UAV overview ................................. 9 2.6 Interesting case study: PPDS Corvo ................................ 11 3 Methodology 13 3.1 Aircraft construction ......................................... 13 3.2 Electronics installation ........................................ 15 3.3 Flight testing ............................................. 15 4 Airplane design and construction 16 4.1 Conceptual design .......................................... 17 4.1.1 Top Level Requirements ................................... 17 4.1.2 Initial design configuration ................................. 18 4.1.3 Battery capabilities and energy storage .......................... 19 4.2 Preliminary Design .......................................... 20 II CONTENTS CONTENTS 4.2.1 Wing loading design ..................................... 21 4.2.2 Weight and wing surface estimation ............................ 22 4.2.3 Fuselage design ........................................ 23 4.2.4 Airfoil selection ........................................ 24 4.2.5 Wing design .......................................... 26 4.2.6 Horizontal tail and longitudinal stability .......................... 32 4.2.7 Vertical tail and lateral stability .............................. 36 4.2.8 Control Surfaces ....................................... 38 4.2.9 Landing gear ......................................... 39 4.2.10 Power plant and energy management ............................ 39 4.2.11 Weight centering ....................................... 41 4.3 Detailed design ............................................ 42 4.3.1 Fuselage module ....................................... 42 4.3.2 Wing module ......................................... 43 4.3.3 Tail module .......................................... 44 4.3.4 Full airplane 3D model .................................... 45 4.4 Building process ........................................... 45 4.4.1 Fuselaje ............................................ 46 4.4.2 Wing .............................................. 48 4.4.3 Tail .............................................. 50 4.4.4 Conclusion of the building process ............................. 51 5 Electronic systems 52 5.1 Basic electronics: propulsion plant and actuators ......................... 52 5.2 Autonomous Flight Controller (FC) ................................ 53 5.2.1 PixHawk board ........................................ 54 5.3 Sensors ................................................ 54 5.4 Ruby FPV: data transmission system ............................... 56 5.4.1 Ruby FPV hardware setup ................................. 57 5.5 Vehicle electronics layout ...................................... 59 5.6 Ground Control Station ....................................... 60 5.6.1 Mission Planner GCS software ............................... 60 5.6.2 Ground Control Station hardware layout .......................... 61 6 Flight tests 62 6.1 Flight I: first take off ......................................... 62 6.2 Flight II: electronics onboard .................................... 62 6.3 Flight III: crash ............................................ 63 6.4 Flight IV: flight modes ........................................ 64 6.5 Flight V ................................................ 67 III CONTENTS CONTENTS 6.6 Flight tests conclusions ....................................... 68 7 Budget summary 69 8 Conclusions 70 IV T´ıtol del TFE List of figures 2.1 U.S. commercial drone market expected growth 2020-2030. The market size of the three main aircraft configurations is shown. Extracted from [10]. ....................... 6 2.2 Corvo PPDS UAV on the launch catapult. [20]. .......................... 12 2.3 Corvo PPDS UAV crashed on the ground. LiPo batteries (1), Pixhawk Cube Orange (2) and the GNNS antenna (3) can be distinguished. [21]. ......................... 12 4.1 Wing loading and aspect ratio parameters for five model types. The power loading column is ignored as it refers to internal combustion engines. Extracted from [25]............. 21 4.2 Cruise speed vs wing loading for different CL values, from Equation 4.1. Extracted from [25, Chapter 4]. .............................................. 21 4.3 calculation procedure of acceptable aero-surfaces. ......................... 23 4.4 Fuselage prototype, designed around the components that have to be placed inside. ...... 24 4.5 Compared airfoils. Plotted by Airfoil Tools [27]. .......................... 25 4.6 Wind tunnel tests results of the Clark-Y airfoil for various Reynolds numbers. Extracted from [26, Appendix 2]. ........................................... 26 4.7 Airfoil approach for foam board wing construction. ........................ 26 4.8 How Aspect Ratio (AR) affects lift coefficient slope CLα. Extracted from [25, Chapter 1]. . . 27 4.9 Compared wing geometries. The top side of the geometries corresponds to the leading edge. Plotted on XFLR5 [28]. ....................................... 28 4.10 Simulation results for the four compared wing geometries, simulated at design cruise speed of 13.4m/s (48km/h). The chosen solution is highlighted in green. Type 1 analysis: constant speed. Analysis method: Horseshoe vortex (VLM1). ....................... 29 4.11 Stall progression patterns for various planform wings. Extracted from [25, Chapter 5]. . . . . 31 4.12 Common dihedral angles for three wing vertical positions. Extracted from [25, Chapter 9]. . . 31 4.13 How parasitic drag and induced drag contribute to total drag in terms of velocity. Extracted from [25, Chapter 5]. ......................................... 31 4.14 Simulated stream after the aerodynamic surfaces. Downstream effect can be appreciated, while the stabilizer remains below the wake. Simulated on XFLR5. .................. 33 4.15 . Extracted from [26, Chapter 12]. ................................. 33 V LIST OF FIGURES LIST OF FIGURES 4.16 Relative pitch stability for various locations of CG along the MAC . Extracted from [25, Chapter 6]. .............................................. 34 4.17 How tail incidence affects airplane’s moment coefficient. CG on 30% MAC . Green plot corresponds to stabilizer incidence of -0.2 degrees, red for -2 degrees. Plotted on XFLR5 [28]36 4.18 CAD model drawn on Solid Works. 1:1 scale. ........................... 37 4.19 Area rule application to determine the vertical tail surface and geometry. The three vertical lines correspond to 25, 30 and 35% of the TMA. ......................... 38 4.20 Constructive parameters on the landing gear assembly. Extracted from [25, p. 75]. ...... 39 4.21 Manufacturer specifications for D3536-8 1000KV electric motor. ................. 40 4.22 3D views of the different fuselage parts. .............................. 43 4.23 Detail of motor mounting incidence: 2◦down, 2◦right. ...................... 43 4.24 Different views on the three wing detachable modules. ...................... 44 4.25 Tail module views. .......................................... 44 4.26 Fuselage - tail union. The transparent pieces are attached to the fuselage, while the opaque ones are part of the tail. ....................................... 45 4.27 Full plane assembly .......................................... 45 4.28 Finished 3D printed fuselage parts. The different arrangements of the parts on the print bed and the support material can be seen. ............................... 46 4.29 Unfolded fuselage views. ....................................... 47 4.30 Views of the two versions of the fuselage. The second version is at the bottom of the images. 47 4.31 Detail on left side of the central wing section. ........................... 48 4.32 Unfolded wing sections. ....................................... 48 4.33 Here the side sections of the wing are joined to the centre section. Once the joints are fixed in place, the side sections are closed. .................................. 49 4.34 Wing - fuselage connection system used on the first version of the fuselage. .......... 49 4.35 Due to the flexing of the wing under the weight of the aircraft, a 5 degree polyhedral ends up being effective in flight. ....................................... 50 4.36 Rudder actuator mechanism. All the actuators operate in the same way. ............ 50 5.1 Pixhawk 1 flight controller board, together with description of its ports and connectors. Extracted form [24]. ......................................... 54 5.2 Steps to configure relay radio link. Extracted from Ruby FPV website [9]........... 57 5.3 Photo of Ruby FPV ground controller hardware layout. HDMI cable is not connected to the Raspberry Pi for aesthetic reasons. RC controller USB cable is neither connected. ...... 58 5.4 Photo of the electronic components of the vehicle mounted in the fuselage. ........... 59 5.5 Ground Control Station hardware layout. ............................. 61 5.6 Ground Control Station before Test Flight IV. .......................... 61 5.7 Full UAS,GCS and UAV, ready to lift off on Test Flight V. ................... 61 VI LIST OF FIGURES LIST OF FIGURES 6.1 Test Flight II relevant data. Obtained from Pixhawk data logs and accessed through Mission Planner. ................................................ 64 6.2 Autonomous mission to be accomplished on Test Flight VI. Configured on Mission Planner. . 65 6.3 Autonomous mission path along with relevant sensor data showing the behaviour of the aircraft. 66 6.4 Altitude, throttle, battery voltage, battery current, airspeed, and angle of attack (Des Pitch) on a level flight path. Flight mode: FBWB. ............................ 66 6.5 Flitht path of Test Flight V. Waypoints are displayed as the previous flight’s mission has not been erased from Pixhawk before the flight. ............................ 67 VII T´ıtol del TFE List of tables 2.1 UAV classification based on weight, operational altitude, range and payload. Extracted from [18]. .................................................. 9 2.2 Reference UAVs. Company, model and aircraft configuration details. .............. 10 2.3 Reference UAV s. Operational capabilities and key design parameters. Bramor ppX (10) is not included in this table. ........................................ 10 2.4 Corvo PPDS operational capabilities and key design parameters.[19]. .............. 12 4.1 Potential aircraft configurations comparison. Five potential configurations are outlined along with their main pros and cons. ................................... 19 4.2 First estimate of total mass and wing area. ............................ 23 4.3 Key parameters of the four airfoil options. Simulation parameters for aerodynamic coefficients: Re=200000, Ncrit=9. Values provided by Airfoil Tools[27]. ................... 25 4.4 Weighted decision matrix. Gain score is ranked from 1 to 4, from the lowest to the highest desirable value. The best option is Clark-Y airfoil. ........................ 25 4.5 Adequate AR for a desired flight behaviour. Extracted from [25, Chapter 5]. .......... 28 4.6 Main geometrical parameters of the considered wing planforms. ................. 29 4.7 Second estimate of total mass and wing area, based on the updated weight of the central wing section. From now on the wing area is fixed. ............................ 30 4.8 Final main wing parameters. .................................... 32 4.9 Neutral Point resulting from Eq. 4.14. Static Margin is reduced as the CG is pushed aft. . . 35 4.10 ..................................................... 36 4.11 Final design parameters for vertical tail plane. .......................... 38 4.12 Design values for the control surfaces, following. Cells filled with “-” correspond to non-interest values. ................................................. 38 4.13 Main parameters of the final aircraft, once built. ......................... 42 5.1 Sensors fitted to the aircraft, together with the module to which they belong and their technical specification. ............................................. 55 5.2 Ruby FPV hardware list and specifications. ............................ 58 VIII 1.4. JUSTIFICATION CHAPTER 1. INTRODUCTION $1 billion in 2017 in the United States market. This growth is mainly due to the fact that UAV technology is offering a much more cost-effective solution to the markets it is addressing. An article published on the website of the Chinese company JOUAV, claims a 90 per cent reduction on pipeline maintenance time and maintenance costs by using its JOUAV CW-30D hybrid Vertical Take Off and Landing (VTOL)UAV [4]. Popular Unmanned Aerial System ( UAS ) applications include aerial photography and cinematography, monitoring, surveillance, package delivery precision agriculture or 3D mapping [5]. In order to accomplish such missions, two main types of UAV platforms can be distinguished: fixed wing and rotatory wing. Rotary-wing vehicles are noted for their high maneuverability, vertical take off and landing and hover capability. However, their inefficient energy management, inherent in the operational nature of the system, results in poor performance in terms of endurance and range [2]. Fixed wing configuration provides a more efficient flight, resulting in much higher endurance and range capabilities, as well as the ability to lift higher payloads. However, they need a certain airspeed in order to produce lift, what imposes the need of a runway for takeoff and landing [2]. Fixed-wing VTOL UAV s were developed to integrate the advantages of both rotatory and fixed wing. Assuming a higher operational load compared to a pure fixed wing aircraft, these hybrids are capable of taking off and landing almost anywhere, while maintaining competitive fixed-wing performance on cruise flight. Deployment systems, such catapults or catch nets, are also commonly used to overcome the needing of a runway on fixed-wing UAV s, thus expanding the possible scenarios for the operation of these aircraft. The UAV s Zipline Platform 1 [6] and Bramor ppX [7] offer an interesting exemplification of this deployment systems, which are further explained in Section 2.5. The progress in electrification has resulted in a broader adoption of electric propulsion systems in these unmanned aircraft. Electric power plants are simpler, more efficient, require less maintenance and feature higher operational capabilities than an internal combustion one. However, electric propulsion is limited by the low energy density of batteries, around 1000 times lower than gasoline [8], which makes electric aircraft feasible only for missions where their short range is not a problem. Reference aircraft, with electric power plant and similar characteristics to the one developed in this work, account for flight times around 60 minutes (Table 2.3), value that meets the requirements of the mission to be accomplished. Thus, the lack of endurance due to battery low energy density will not be a problem in this case. That said, since a tight budget and simplicity and practicality in design are fundamental requirements in the developed aircraft, a fixed-wing configuration with electric propulsion is chosen, which is intended to have Short Take Off and Landing (STOL) capabilities. The avionics a crucial aspect on the development and evolution of UAS technology. Here a distinction between three main systems can be made: the energy management unit, the communications system and the autopilot. The power management unit is responsible for distributing power from the batteries to the motors and actuators, as well as adapting and filtering the power supply to the on-board electronics. The communication system is in charge of receive the control orders from the Ground Control Station ( GCS ) and send status information of the aircraft, as well as relevant data such as payload status or HD video feed. 3 1.4. JUSTIFICATION CHAPTER 1. INTRODUCTION In this project, the communication system will be undertaken by Ruby FPV [9], an open source complete platform (software and hardware) designed for controlling and managing UAV s with robust end to end digital radio links; running on Raspberry Pi microcomputers. Ruby FPV has been selected for integrating in the same platform the telemetry link, control, HD video feed, and the ability to act as a relay node to extend the range of a second vehicle; all managed through the same platform. The autopilot system plays a vital role in controlling and maneuvering the UAV , ensuring its stability, navigation, and adherence to flight plans. It is capable of autonomously make decisions and execute commands based on pre-programmed instructions and real-time sensor data. By automating flight operations, autopilots enhance safety, efficiency, and accuracy in various applications. Due to increasing popularity on UAV industry, both civilian and military companies have developed their own solutions to overcome the control of their vehicles. Thus, at the commercial level, there can be found both aircraft that integrate an in-house built autopilot, such as drones made by DJI and Autel Robotics, as well as autopilot systems that are ready and approved for installation in your aircraft, as the ones developed by UAV Navigation or Cube Pilot. Despite of the great variety, the focus will be on the open source ArduPilot project [1], which was officially launched by Chris Anderson and Jordi Mu˜noz in 2009. ArduPilot is an open source, unmanned vehicle autopilot software suite, capable of controlling a wide variety of autonomous vehicles [1]. Initially created by enthusiasts with the purpose of governing model aircraft and rovers, ArduPilot has evolved into a robust and full-featured autopilot system that garners trust from industries, research institutions, and hobbyists alike. The ArduPilot software suite comprises of navigation software installed on the vehicle, as well as Ground Control Station ( GCS ) software such as Mission Planner, APM Planner, QGroundControl, MavProxy, and various other tools [1]. ArduPilot suite is nowadays compatible with a wide list of embedded hardware, which connects to peripheral sensors and devices employed for navigation. Pixhawk flight controller boards are the hardware platform for Ardupilot software. Pixhawk boards are compatible with UART, I2C, CAN and SPI serial communication protocols, enabling efficient data exchange with other on-board computers, such as Raspberry Pi or Jetson nano, allowing implementations with computer vision, AI, advanced navigation, or complex algorithm processing functions, among many others. Having introduced the UAV technology, its projection in today’s market and the capabilities of the ArduPilot and Ruby FPV projects, this work is based on the idea to combine the lastest open source technology available on the fixed-wing UAV field, in order to create a prototype of autonomous aircraft. The prototype will be the first iteration on the design and build of a low cost a platform, built with cheap and accessible materials, capable of, while flying autonomously, acting as relay node between a Ground Control Station ( GCS ) and an other UAS , enabling the control of the last one BVLOS . This system will have clear applications in search an rescue missions, where the drone operating BVLOS will offer an extended range and closer inspection of the terrain, without the need of ground infrastructure beyond GCS , featuring thus low deployment times, flexibility, and operational simplicity. 4 T´ıtol del TFE Chapter 2 State of the art As it has been introduced in the previous Chapter, the civilian UAV market has experienced a major growth over the last two decades, and it is expected to continuously grow over the next years. The market growth is attributed to the increasing enterprise application of drones across diverse industry sectors. Multiple drone manufacturers are continuously engaged in testing, creating, and improving solutions for a wide range of markets, giving as a result a substantial growth of the business applications of commercial drones in recent years. Additionally, the integration of state-of-the-art technologies into commercial drones such as precise sensors, high resolution cameras, more powerful on-board processing units, Artificial Inteligence ( AI ), or Machine Learning ( ML ) capabilities has lead to deliver improved solutions, and is creating new opportunities for growth within the commercial drone industry. 2.1 Market overview According to a market analysis conducted by Grand View Research [10], based on historical data between 2018 and 2021, the global UAV market size value is estimated on 22.98 billion USD in 2023, and is expected to grow at a Compound Annual Growth Rate (CAGR) of 13.9% from 2023 to 2030. This value of CAGR is close to what is considered a healthy range of 8% to 12%, for companies who have been around for 10 or more years, or 10% to 20% for smaller companies of start-up business [11]. A healthy CAGR implies constant growth over time, long-term viability, market stability and investor confidence. Also The Bussiness Research Company [12] reports a global market value of 26.03 billion USD and expects it to grow to $52.21 billion on 2027, at a CAGR of 14.9%. G. V. Research also provides an estimation of the US commercial drone market growth between 2020 and 2030, displayed in Figure 2.1. Here, a distinction is made between the three main UAV aircraft configurations: fixed wing, rotatory blade, and hybrid (also called VTOL ). Every configuration has its pros and drawbacks, and is suitable for a particular range of applications. These pros and cons are discussed further in Justification section 1.4. 5 2.2. BATTERY-POWERED UAV!s CHAPTER 2. STATE OF THE ART Figure 2.1: U.S. commercial drone market expected growth 2020-2030. The market size of the three main aircraft configurations is shown. Extracted from [10]. 2.2 Battery-powered UAVs The advances on electrification have lead to a wider use of the electric propulsion plant on the UAV sector. Nowadays electric motors have demonstrated that can compete with the internal combustion ones, offering a similar or even greater power performance [8, p.736] while increasing energy efficiency, simplicity and reliability [8, p.739-740]. In addition, electric motors are the only possible option on multicopter vehicles, where the control is achieved through differential motor speed drive; as internal combustion engines cannot be driven at such high frequencies. The main drawback of the electric propulsion is the energy storage. Whereas the energy density of the gasoline is around 12000 Wh/kg , batteries currently available on the market are only reaching 200 Wh/kg (18650-type cylindrical Li-ion), being this the reason why the endurance of most of the electric-powered UAS in the market is between 60 and 120 min, as can be seen in Table 2.3. Because it is such a critical aspect of transport electrification, a lot of resources are being invested by companies and academia to obtain batteries with higher capabilities. Thus, while the energy density on commercially available lithium batteries oscillates around 140 Wh/kg (Tattu 3S 2200mAh LiPo) for LiPo batteries and 200 Wh/kg (18650-type cylindrical Li-ion) for Li-Ion batteries, those developed at industrial level by manufacturers such as Tesla are reaching values of 272 − 296 Wh/kg [13](4680-type cylindrical Li-ion). On March 2023 researchers from Tianjin University, China, managed to develop a pouch cell with an ultrahigh energy density of 521.3Wh/kg, tested at discharge ratios up to 5C [14]. 2.3 Application insights This expected growth of the commercial drone market is entirely due to the wide range of scenarios in which drones offer an alternative that, while providing similar or better results compared to conventional means, represents a significant improvement in terms of time save and cost-effectiveness. Capabilities such as autonomous flight, geo-awareness, AI assisted data processing or high resolution sensors and cameras, led to a wide range of business use cases for commercial drones. Also miniaturization has allowed for the integration 6 2.3. APPLICATION INSIGHTS CHAPTER 2. STATE OF THE ART of powerful sensors, processors, and communication systems into smaller airframes, resulting in UAV s that are not only more compact but also more capable. Some of the key sectors where drones are being used are listed below. 2.3.1 Aerial photography and videography The media and entertainment sector held a market portion exceeding 22% in 2022, and it is projected to experience significant expansion throughout the 2020s. [10]. Multirotor configurations are common here for their wide operational flexibility, with DJI being one of the world leaders on designing and manufacturing of those videography unmanned aircraft, including features such as great endurance, autonomous flight, 360º space vision, or AI assisted high resolution filming cameras. First Person View ( FPV ) drones are also playing an important role in the film and cinema industry, taking existing cameras through high-speed and acrobatic aerial manoeuvres, and achieving shoots that where never seen before. Finally, drones such Alta X, designed and manufactured by Freefly Systems, are conceived for lifting heavy high-end cinema rigs, with a MTOW of 34.86kg and a maximum PL weight of 15.06kg. 2.3.2 Precision agriculture Drones employed on precision agriculture allow farmers to monitor crops, assess plant health, and apply pesticides or fertilizers more efficiently. Numerous companies are offering their integrated hardware and software solutions for the mapping and monitoring tasks. Aircraft configurations are going from easy-tooperate < 5kg multicopters, up to high endurance long range fixed wings, passing through a sort of hybrid VTOL configurations. Those are equipped with payloads such as multispectral cameras, thermal and RGB cameras, LIDAR sensors or precision spraying systems, among others. Some of the companies offering mapping solutions are DeltaQuad with its DeltaQad PRO VTOL mapping UAV , Parrot with its Parrot Bluegrass Fields quadcopter, or DJI with its Mavic 3 Multispectral. For fertiliser spraying and spreading, the DJI AGRAS T40 stands out, capable of carrying up to 50kg of spreader payload, as well as having topography and mapping capabilities. 2.3.3 Surveying, mapping and inspection Multicopters can easily access to hard-to-reach areas, making them valuable for inspecting infrastructure such as bridges, buildings or electricity pylons, enhancing safety and reducing costs. UAV s on their different configurations (depending on the requirements of the mission) are utilized to create accurate 3D maps and survey large areas quickly, as well as monitoring and securing large events, critical infrastructure, and remote locations. Furthermore, drones have a vital function in disaster response by offering immediate situational understanding and assisting in search and rescue missions. Notable here is the DJI Matrice 200 V2 series, capable of switching between six different payload cameras, third-party payloads aside. Autel Robotics also manufactures specialised UAV with its EVO series (multi-task quadcopters) and its DragonFish series, a fixed wing hybrid VTOL UAV , capable of lifting five different specifically designed payload cameras, offering long-range capabilities and a great operational flexibility. 7 2.4. CLASSIFICATION AND REGULATION OF UAV!s (UAV!s)CHAPTER 2. STATE OF THE ART Finally, fixed wing UAV are suited for missions requiring the highest possible endurance while lifting heavy payloads. It is also worth mentioning drones like Delair UX11, Precision Hawk Lancaster 5 or senseFly eBee; which stand for light weight drones equiped with specialized and miniaturized payloads capable of take over mapping and surveying tasks, featuring high operational flexibility due to their reduced size and STOL capabilities. 2.3.4 Delivery and logistics Businesses are actively considering the deployment of drones for the final stretch of delivery, aiming to decrease shipping times and costs. Although companies such as Wing, Zipline or Amazon have taken the first steps in this sector, issues such as current regulations, airspace management, infrastructure requirements, noise pollution, aircraft safety or the profitability of the business mean that this field still requires strong development, linked to an improvement in current technology [15]. Although not ready for large-scale application, this technology has already borne fruit in very specific applications, such as the express delivery of critical goods, including medical supplies. This is the business model of Zipline, which uses the Zipline P1 aircraft to deliver urgent medical supplies to countries that lack good ground transport infrastructure, and where health facilities are unable to maintain an acceptable stock of medicines and supplies. Its work in Ghana [16] and Rwanda [17] stands out. 2.4 Classification and regulation of UAVs The gross of UAV s, both commercial and military, can be classified according to various of their features, such as weight, wing type, propulsion type, or operational capabilities, among others. Different organisations (NASA, FAA in the US; EASA in Europe and AESA in Spain) have defined main UAS categories, most of them based on the above-mentioned characteristics. An example of this classification can be observed in Table 2.1, extracted from Reference [18]. Regarding to this classification, G. V. Research report [10] states that “The < 25 kg segment accounted for the largest market share of 81% in 2022.”, segment which corresponds to Class I on Table 2.1. The report refers to operational flexibility, easy operational functionality, and a wide range of applications while conforming to stringent UAS regulations, as the main reasons of the dominance of Class I UAS in commercial UAV market. 8 2.5. CURRENT AUTONOMOUS UAV! OVERVIEW CHAPTER 2. STATE OF THE ART Category NASA UAS Class Mass [kg] Normal Operating Altitude [m] Mission Radius, Range [km] Typical Endurance [h] Payload [kg] Available UAV Models in Market Micro <2<140 5 <1<1 DJI Spark, DJI Mavic, Parrot Bebop2 Mini sUAS Class I 2-25 <1000 25 2-8 <10 DJI Matrice600, DJI Inspire2, Airborne Vanguard Small 25-50 <1700 50 4-12 <50 AAI Shadow 200 Medium Class II 150-600 <3300 200-500 8-20 <200 Griff 300, Ehang 216 Large/ Tactical Class III >600 >3300 >1000 >20 >200 Boeing X-45A UCAV Table 2.1: UAV classification based on weight, operational altitude, range and payload. Extracted from [18]. 2.4.1 Drone regulations in Spain The drone regulations by AESA (Agencia Estatal de Seguridad A´erea) in Spain have undergone updates and adjustments, aligning with the European regulations established in Delegated Regulation (EU) 2019/945, which came into effect on December 31, 2020. These regulations classify drones into different categories based on their risk level. It defines six classes of drones, from C0 to C6, which can operate in three categories: open, specific, and certified. Pilot certification and drone licenses are common requirements across all classes. According to European regulations, from January 1, 2024, drones used in the open category must belong to one of the classes C0, C1, C2, C3, or C4, and drones operating in standard scenarios ’STS-01’ and ’STS-02’ must have a C5 or C6 class identification label. Additionally, they must have identification labels for operations in European standard scenarios. Drones introduced into the European Union before January 1, 2024, can continue flying under open subcategory A3 (away from populated areas and people). Frequencies used and power levels used on the radio link must respect the restrictions imposed by European regulations, which require the use of an LBT (Listen Before Talk) protocol. Regarding airspace, in Spain, it is structured within the AIP, and divided into controlled and uncontrolled airspace. ENAIRE, air traffic management and air navigation services provider in Spain, offers the ENAIRE Drones web service. This service encompasses all relevant information about flight zones in Spain, as well as coordination mechanisms for drone operations in restricted airspace, nature protection areas, and standard scenarios (STS). 2.5 Current autonomous UAV overview The specifications and operational capabilities of the solution developed in this work correspond to this Class I, specifically to the Micro ( < 2kg) category, specified in Table 2.1. Therefore, eight Class I Mini UAV s have been selected as reference models for establishing a guide to the most common aircraft configurations for each application, as well as providing a clearer picture of the capabilities and performance of the aircraft that constitute today’s market standard, in the field of fixed wing UAV. 9 2.5. CURRENT AUTONOMOUS UAV! OVERVIEW CHAPTER 2. STATE OF THE ART That said, the technical specifications and operational capabilities of the aforementioned drones are set out in Tables 2.2 and 2.3 below. Company Model Ref. num. Engine Wing type Motor config Application CAT UAV ORYX 1 4 strokes, EFI Rectangular Tractor Great endurance multitask CAT UAV ATMOS-8 2 Electric Elliptical Pusher Utility CAT UAV EXO C2-L 3 Electric Elliptical Tractor Utility ZIPLINE Zipline 4 Electric Elliptical Pylon. 2 coaxial Delivery DELAIR UX11 5 Electric Flying wing Pusher Mapping JOUAV CW-007 II 6 Electric + VTOL Regular + Hquad Pusher + H quad Utility Autel Robotics Dragonfish 7 Electric + VTOL Rectangular + +quad Bimotor + + quad AI camera AVEM UAV AVEM 215 8 Electric Elliptical high \ac{AR}Tractor Mapping AVEM UAV AVEM 400 9 Electric Elliptical high \ac{AR}Tractor Mapping C-ASTRAL Bramor ppX 10 Electric Flying wing Pusher Mapping Table 2.2: Reference UAVs. Company, model and aircraft configuration details. Ref. num. Wingspan [m] MTOW [kg] WL [kg/m2] PL [kg] Endurance Cruise Speed [km/h] 13 22 NA 7.5 26 h 72 2 1.98 3.3 11.79 0.8 100 min 68 3 1.58 1.35 NA 0.25 75 min 65 4 3.7 20 21.51 1.75 170 min 101 5 1.2 1.6 NA NA 60 min 54 62.2 6.8 16.59 0.8 60 min 68.4 7 2.3 9 21.28 1.5 126 min max. 108 8 2.15 2 6.13 0.5 2h 60 9 4 8 8.63 2.3 5h 75 Table 2.3: Reference UAV s. Operational capabilities and key design parameters. Bramor ppX (10) is not included in this table. As can be seen in Table 2.2, the vast majority of the considered UAV s are equipped with electric propulsion systems. This is because they are tailored for mission profiles specifically designed to match the autonomy offered by current battery technology, despite its lower energy density. Therefore, the maximum flight times of the considered aircraft range from 60 to 170 minutes, a result that does not compare with the 26 hours offered by the Cat UAV ORYX (1) model, which features an internal combustion engine. It’s also worth mentioning the 5-hour autonomy of AVEM 400 (9), highlighting the performance advantage of fixed-wing configurations over fixed-wing VTOL (6 and 7), which, with similar MTOW , offer a fraction of the autonomy. The main advantage of VTOL aircraft is their operational flexibility, as they require very small spaces for 10 2.6. INTERESTING CASE STUDY: PPDS CORVO CHAPTER 2. STATE OF THE ART takeoff and landing. In contrast, their flight efficiency and autonomy are compromised due to a higher operational load and the presence of structures that generate additional aerodynamic drag. Nevertheless, designs like Dragonfish (7) or Transwing, by Petro Dynamics, offer intelligent solutions to minimize these drawbacks. Pure fixed-wing models often use catapults for takeoff to become airborne. Regarding landing, Zipline (4) employs a mid-air capture mechanism, while Bramor ppX (10) uses a parachute. In the case of light aircraft, hand launch is common for takeoff, and they simply reduce their speed on the ground for landing, as seen in UX11 (5) or AVEM 215 (8). Net recovery is also a common landing mechanism. Key design parameters of the reference UAV s have been compiled in Table 2.3. These parameters have been primarily obtained from the manufacturer’s specifications, and offer a more developed view of the performance of drones that are the current market standard. The estimation of wing area for determining wing loading has been conducted using the plan view and wingspan provided by the manufacturer, creating a wing planform model in Solid Works to calculate its area. It’s worth noting that VTOL drones have the highest wing loading due to their higher Operational Empty Weight. Zipline (4) also has the highest wing loading, which is logical considering it’s a cargo transport drone. It’s important to consider that it takes off using a pneumatic catapult, that accelerates it to its cruise speed, and lands with a complex system that captures the aircraft mid-air, all at a specifically constructed base of operations. Therefore, it’s evident that as aircraft performance increases, operational flexibility is reduced. Aircraft with lower wing loading and higher autonomy (compared to aircraft of the same wingspan) are AVEM 215 (8) and 400 (9). Their mission involves mapping large areas, requiring slow, stable, and highly efficient flight; while carrying a payload specifically designed for the task. With a 2.15m wingspan, a 2kg MTOW , and a wing loading of 6.13m, the AVEM 215 (8) is the closest reference UAV to the one designed in this work. Therefore, it will be taken as the competitive industrial UAV reference that the designed aircraft should match in order to compete in the current industrial fixed-wing drone market. 2.6 Interesting case study: PPDS Corvo This project aims to construct a UAV prototype controlled by Ardupilot using simple construction techniques and affordable materials. A similar paradigm can be seen in Corvo PPDS (Precision Payload Delivery System) UAV , developed by the Australian consultancy SYPAQ Systems in collaboration with the Australian Defense Forces Academy. With a focus on the military sector, this project has created a lightweight aircraft with minimal production cost and complexity capable of a single flight carrying several kilograms of payload. Thus, this UAS provides transportation and aerial attack capabilities on the battlefield through a low-cost expendable vehicle. The UAV , whose design is rooted in model aircraft, consists of a fuselage box and a wing in a tail-less configuration, held together by elastic rubber bands. It is constructed from waxed foam board and is delivered in flatpack form as an assembly kit [19]. This kit includes pre-cut fuselage and wing sheets, as well as the required electronics, assembly tools, and a tablet for loading the mission flight path. 11 2.6. INTERESTING CASE STUDY: PPDS CORVO CHAPTER 2. STATE OF THE ART Figure 2.2: Corvo PPDS UAV on the launch catapult. [20]. Figure 2.3: Corvo PPDS UAV crashed on the ground. LiPo batteries (1), Pixhawk Cube Orange (2) and the GNNS antenna (3) can be distinguished. [21]. Regarding its navigation systems, Corvo PPDS only has an autopilot and flies according to a precompiled program, without the capacity communicating with a GCS while on flight. Both the hardware and software it employs are essentially the same as those to be used in this project. The flight controller used is the Pixhawk Cube Orange (Figure 2.3), manufactured by Cube Pilot, which, along with a GNNS antenna and a wind speed sensor, provides its autonomous flight capabilities. The software is based on the Ardupilot software suite. Its propulsion system is electric, using high-capacity LiPo batteries (Figure 2.3)that provide an autonomy of between 1 and 3 hours [19]. Wingspan [m] MTOW [kg] WL [kg/m2] PL[kg] Endurance Cruise Speed [km/h] 2 5.4 6.04 3 1 to 3 h 60 Table 2.4: Corvo PPDS operational capabilities and key design parameters.[19]. The drone was supplied to Ukraine in May 2023, and according to Ukrainian media reports, 16 of these units were deployed on August 27, 2023, in an attack on the Dursk Vostochny Airport in Russia [21]. Although its effectiveness in this mission has been questioned, the existence of this UAV , built in inexpensive materials and cheap available software and hardware, highlights the democratization of this technology, as well as the fruits of the efforts and collaboration of the community of enthusiasts and scientists on the Ardupilot suite under the open source licence. 12 4.1. CONCEPTUAL DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION Aircraft configuration Advantages Disadvantages Conventional - Most tested configuration. - Allows wide range of speeds. - Simple tail construction. - Propeller in the front of the plane does show up on the camera feed. Conventional + twin-engine - Clean camera feed, since propellers are located on the wing. - Added weight due to mounting 2x motors, 2x propellers and 2x ESCs. - Higher power consumption and lower power-to-thrust efficiency (smaller motors are less efficient). Conventional + V tail - Less tail area is required, meaning reduced tail weight and drag. - It looks great. - Complex control system. - Limited yaw control. - Limited pitch authority, what can be critical in certain flight regimes such as stalls or high-speed dives. Rear fuselage propulsion - Clean camera feed. - Motor slipstream does not affect the fuselage. - Some ”heavy” payload is required in front of the CG to balance the nose-up moment produced by the motor’s weight. - Heavy fuselage to tail structure, as it can not interfere with the propeller’s area. - Propeller’s slipstream is directly interacting with the tail surfaces. Flying wing - It is the configuration with the highest aerodynamic efficiency. - Clean camera feed. - Needs higher flight speeds due to lack of tail surfaces. - Not provided with yaw control. - Complex tuning process. Table 4.1: Potential aircraft configurations comparison. Five potential configurations are outlined along with their main pros and cons. 4.1.3 Battery capabilities and energy storage When it comes to energy storage on electrical powered vehicles, lithium batteries are the best and more widely used option today. Those are offering an energy density of 100 − 265 Wh/kg and, compared to other batteries, feature long cycle life (lithium batteries can endure a large number of charge and discharge cycles), relatively low self-discharging rate, fast charging, and high discharge rate. Despite of those advantages, lithium batteries still cannot compete with the 12000 Wh/kg offered by gasoline [8]. This is why the main 19 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION drawback of electric powered vehicles is energy storage. Among the most common lithium batteries, two types can be distinguished: the lithium-polymer (LiPo) and the lithium-ion (Li-Ion). LiPo batteries are in fact an specific type of lithium-ion battery, which uses a high conductivity semisolid polymer electrolyte instead of a liquid electrolyte. This results in flexible, compact, and flat design, and gives the LiPo battery some operational advantages such higher discharge rates and lower internal resistance, compared to Li-Ion. In the other hand, Li-Ion have a higher energy density and can be discharged safely down to lower voltages. This can be seen when comparing the 140 Wh/kg Tattu 3S 2200mAh LiPo and 200Wh/kg 18650-type cylindrical Li-ion. Safety considerations are a major concerning on lithium batteries. LiPo batteries can be more sensitive to overcharging, over-discharging (voltage operating range between 3.5V and 4.2V), and physical damage due to their flexible and thinner design. Mishandling or improper charging can lead to swelling, overheating, or even fires in extreme cases. Traditional Li-ion batteries have a more rigid structure and might have slightly better resistance to physical damage. However, both LiPo and Li-ion batteries require proper care and charging practices to ensure safety. Regarding to the project needs, as the exact range of power consumption remains unknown until the first prototype is flight tested, LiPo batteries are chosen, as a better discharge performance offers a greater safety margin in case of unexpected peaks of power consumption in test flights. At the time of the development of the project, Ovonic 3S 11.1V 1500mAh LiPo batteries are available with an energy density of 138 , 75 Wh/kg . A study of the power required to ensure a certain flight time is not carried out, as it requires an acceptable estimate of the drag coefficient. This can be obtained as a result of finite element studies or through historical data from similar aircraft. The first method is discarded as being outside the scope of the work, while the second would not provide meaningful results either, as it would not take into account effects such as the rough surface finishes of the aircraft or the approximation of the airfoil by straight sections, both of which are due to the materials and construction methods used. Therefore, for this prototype, two 3s 11.1V 1500 mAh batteries stacked in parallel will be used, adding up to a capacity of 3000 mAh . Later, in flight tests (Chapter 6), the performance of the aircraft will be evaluated and the power needed to meet the mission requirements will be discussed. 4.2 Preliminary Design The preliminary design phase is an early stage in the process of airplane design. Here the overall configuration of the aircraft is established, thus developing a comprehensive design plan by involving a through analysis and evaluation of various design parameters and considerations. Is the step prior to the detailed design phase. The preliminary design phase addresses issues such as aerodynamic analysis, performance estimation, structural sizing, systems integration, design trade-offs and economic cost estimation. 20 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION 4.2.1 Wing loading design The wing loading represents the relationship between the aircraft’s gross weight and it’s wing area. It is essentially the amount of weight supported by each wing area unit. This parameter will affect the overall behaviour in the air, and therefore is chosen according to the desired flight performance of the aircraft. Figure 4.1, extracted from reference [25, p. 23], shows five general types of model aircraft together with their most common wing loading and aspect ratio parameters. Figure 4.1: Wing loading and aspect ratio parameters for five model types. The power loading column is ignored as it refers to internal combustion engines. Extracted from [25] Figure 4.2: Cruise speed vs wing loading for different CL values, from Equation 4.1. Extracted from [25, Chapter 4]. The desired behaviour for the designed aircraft is between the soaring glider and the low-speed trainer, hence an initial Wing Loading of 14 oz/ft2 is selected, which is equivalent to 4 . 272 kg/m2 in the International System of Units (SI). Aspect Ratio issues are discussed in Section 4.2.5. In cruise flight conditions lift forces are equalised to weight ( L = W ). Here, the wing loading (W/S), the incoming wind speed (v) and the Lift Coefficient ( CL ), are related through lift force equation, 4.1. Andy Lennon displays graphically this relationship on the nomograph of Figure 4.2, which allows for an approximate but quick calculation of the cruise speed of the airplane with the wing operating at its optimum CL , for a given wing loading. A quick estimate of takeoff and landing speeds can also be found for the given wing loading and the maximum CL which depends on the airfoil and the wing geometry. L=1 2·ρ·V2·CL ·S(4.1) Since the airfoil has not yet been chosen, the optimum CL of the wing is assumed to be between 0.3 and 0.4. Thus, for a wing loading of 14 oz/ft2 on Figure 4.2 the cruise speed will be between 27 and 33 mph , 12.1 - 14.8 m/s , which is consistent. A quick estimate of takeoff and landing 21 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION speeds can also be calculated for the given wing loading and maximum wing CL , which is calculated down in Section 4.2.5. This W/S value is fairly low compared to the ones calculated for the reference UAV s on Table 2.2, this is mainly because those have take-off and landing assistance provided by catapults (take-off), parachutes and catch nets (landing), on the fixed wing models; and VTOL capabilities on the hybrid multirotor-fixed wing models. The developed aircraft is primarily intended to be hand-launched, so a fairly low W/S is needed to sustain the model at such reduced speeds. 4.2.2 Weight and wing surface estimation The gross weight of the UAV has been divided in two major groups in order to carry out a first weight estimation: fixed weight and the variable. The fixed weight is the one that is known at the beginning of the design process, and it is composed of the propulsion plant (its weight is known since the motor to be used is an input parameter), the batteries, the electronics (receiver, servos and BEC) and the payload. The variable weight is the one depending on weight of the built structures. Thus it’s composed of the weight of the wing, the fuselage, the tail and the 3D prints that will hold the parts together. As these structures have not yet been calculated, their weight must be estimated as the design process progresses. The fuselage, wing and tail are made of 5mm thick foam board, which density is 0 . 551 kg/m2 . Thus the weight of the foam board structures can be estimated by knowing their surface. The aim of these section is to estimate the wing area, that is depending on the wing loading and the estimated gross weight of the aircraft. Hence they are related through the following equation: S=W/(WL) (4.2) The fuselage 4.2.3 is designed around the electrical components and payload, and it will not have major changes during the design process. Hence, its weight is fairly well estimated from the starting point. The weight of the tail boom and the 3D prints is also directly estimated. Concerning to wing and tail surfaces, its weight is directly depending on the wing area. Thus, once the relationships between these values have been defined in equations 4.3 and 4.4, a simple iterative procedure have been developed in Matlab to obtain an acceptable estimate of aero-surfaces. The used algorithm is schematized on Figure 4.3, while the code and the results of the initial weight estimation can be found on Annex A. Wwing = (2 ·SW·ρfoamboard)·1.2 (4.3) Wtail = 0.15 ·SW·ρfoamboard (4.4) Wtotal =Wfixed +Wvariable(SW) = Wfixed +Wwing(SW) + Wtail(SW) + Wfuselage +Wtailboom +W3Dprints (4.5) In the equation 4.3, the factor 2 is due to the fact that the wing is made up of two sheets, which constitute 22 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION the inboard and the outboard. A correction factor of an additional 20% is applied. In the equation 4.4, the total tail surfaces are estimated at 15% of the estimated wing area. This value is based in historical data for this specific configuration. In fact, the resulting total tail surfaces (vertical + horizontal) at the end of the design phase is on 16% of the wing area. Figure 4.3: calculation procedure of acceptable aero-surfaces. The following Table 4.2 shows the key design values from this section. Also the values for the fixed, variable and estimated weights can be appreciated on Annex A. Total Mass [kg] Wing Area [m2] Wing Loading [kg/m2] 1.298 0.3038 4.272 Table 4.2: First estimate of total mass and wing area. 4.2.3 Fuselage design A fuselage prototype is the first structure to be designed. This is because this structure is not expected to experience significant changes throughout the design process, so an estimate at this early stage allows for an approximation of its weight and the interior layout of the components, providing an overview of the distribution of these components within the available space. 23 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION The fuselage is therefore 3D modelled around a preliminary disposition of the power plant and the payload parts intended to be placed inside, resulting in the following prototype: (a) Components listed from front to back: motor, PixHawk, ESC , BEC , batteries, Raspberry Pi, WiFi module, and tail boom. (b) Fuselage box prototype. Figure 4.4: Fuselage prototype, designed around the components that have to be placed inside. Minor dimensional changes will be applied to this model to accommodate all components along the design process progresses, as well as to balance the CG . Thus, this prototype looks very similar to the final version and therefore its weight can be reasonably estimated. The final version of the fuselage is displayed in Section 4.3, Detailed design. 4.2.4 Airfoil selection Chapter 2 on Lennon’s book [25] states that “The selection of an airfoil section for most powered models is considered not to be critical [...]. Models fly reasonably well with any old airfoil [...]. In contrast, the R/C soaring fraternity is very conscious of the need of efficient airfoils.”. In addition, the materials chosen and the construction techniques used do not make it possible to reproduce the exact airfoil of the wing, but rather to build an approximation of it, as it can be seen in Figure 4.7. That said, an in-depth study of the optimum airfoil is not carried out, but the airfoil chosen is selected from four suitable options, displayed on Figure 4.5. For these, the key performance values have been compiled using the data available in Airfoil Tools [27] (Table 4.3). Then these data has been ranked using a weighted decision matrix (Table 4.4), which provides an absolute index of the feasibility of each option. A most important consideration in airfoil selection and wing planform design is “scale effect”. The measure of scale effect is Reynolds number (Re), defined as: Re =ρ·v·c µ(4.6) Being ρ the air density, v the fluid velocity, c the airfoil chord, and µ the fluid dynamic viscosity; all magnitudes in SI. To have an example, a full-scale airplane flying at 83m/s (300km/h) with a wing chord of 1.5m is operating at Re 8,400,000 (considering ISA conditions at sea level), while a scale model flying at 16.6m/s (60km/h) with a wing chord of 0.2m flies at Re 220,000. This value is reduced when landing at lower speeds. In Figure 4.6, extracted from M. Simons’ book [26], wind tunnel tests over a wide range of Reynolds number reveal the adverse impact low Re flight on airfoil performance. 24 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION This is why “low Re ” airfoils are used on model airplanes, offering slightly better performance on the low Re operating range. All four airfoils presented here feature this “low Re” performance. (a) Eppler E197 (b) Eppler E374 (c) Selig S7055 (d) Clark-Y Figure 4.5: Compared airfoils. Plotted by Airfoil Tools [27]. Airfoil Max. Thickness % α Stall Clmax Max. Cl/Cd E197 13.49 10 1.19 78 E374 10.91 11.7 1.15 75 Selig S7055 10.5 15 1.26 77 Clark-Y 11.7 15.7 1.34 74 Table 4.3: Key parameters of the four airfoil options. Simulation parameters for aerodynamic coefficients: Re=200000, Ncrit=9. Values provided by Airfoil Tools[27]. Criteria Weight E197 E374 S7055 Clark-Y GW·GGW·GGW·GGW·G Max. Thick. % 1 4 4 2 2 1 1 3 3 α stall 2 1 2 2 4 3 6 4 8 Clmax 3 2 6 1 3 3 9 4 12 Max. Cl/Cd 3 4 12 2 6 3 9 1 3 24 15 25 26 Table 4.4: Weighted decision matrix. Gain score is ranked from 1 to 4, from the lowest to the highest desirable value. The best option is Clark-Y airfoil. Clark-Y airfoil proves to be the best option among those considered. It features the greatest stall AoA and 25 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION Clmax , while still having the second highest maximum thickness. More maximum thickness allows for more space for structural reinforcement and actuators installation. In terms of aerodynamic efficiency, although it has the lowest of the four, the difference in performance compared to the E197 airfoil, which has the highest value, is small. Furthermore Clark-Y features a flat bottom shape, which facilitates the construction process, and a gentle stall pattern that allows the aircraft to still have some manoeuvrability at high angles of attack. Figure 4.6: Wind tunnel tests results of the Clark-Y airfoil for various Reynolds numbers. Extracted from [26, Appendix 2]. Figure 4.7: Airfoil approach for foam board wing construction. 4.2.5 Wing design In order to find an appropriate wing design for the desired airplane type and flight behaviour, decisions must be made on the following design variables: •Airfoil selection (undertaken in previous paragraph). •Wing planform. •Aspect ratio. •Stall patterns, lift distribution and stall avoidance. •Wingtip design. •Flaps. 26 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION Wing planfrom design, along with stability analysis, has been done with the help of XFLR5 software [28]. As they state in their website, XFLR5 is an analysis tool for airfoils, wings and planes operating at low Reynolds Numbers. Aspect Ratio As explained in [25, Chapter 5], the Aspect Ratio ( AR ) of a wing has a major impact on its induced drag, defined as that drag caused by the development of lift. The classical formula describing drag coefficient is shown in Equation 4.7. The first term of the equation refers to parasitic drag, and the second to induced drag. It can be observed that both the Oswald factor ‘ e ’, which also depends directly on the AR , as well as AR itself, contribute directly to reducing this induced drag. Then so, a higher aspect ratio reduces induced drag, as narrower chord tips result in smaller wing tip vortices. Also the lift per AoA increases, as it can be seen in Figure 4.8, resulting in higher lift produced at a lower AoA. For a givenAR and wing planform, this slope can be calculated through Equation 4.8. CD=CD0+C2 L π·e·AR (4.7) dCL dα ≡a=a0 1 + a0 πAR ·(1 + τ)(4.8) e= 1.78 1−0.045AR0.68−0.64 (4.9) •a= actual lift slope of the wing. •a0= lift slope for an infinite wing, often a0= 2π. •τ= tabulated planform adjustment factor. •e = Oswald factor, empirically approximated in 4.9 for wings with leading edges with no sweepback. Proposed by [29]. Figure 4.8: How Aspect Ratio (AR) affects lift coefficient slope CLα. Extracted from [25, Chapter 1]. All this favours the use of higher AR . In contrast, wings with higher AR are less manoeuvrable and have 27 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION greater structural complexity (longer spans increase bending moments on the root and narrower chords reduce space for internal structural reinforcement). That is why a concrete range of AR is suitable for every desired flight behaviour, as can be seen on Table 4.5. Model type Aspect Ratio High-speed, highly maneuverable 4 to 6 Moderate-speed sport 6 to 8 Low-speed trainer 8 to 10 Slope gliders 8 to 10 Soaring gliders 10 to 15 Table 4.5: Adequate AR for a desired flight behaviour. Extracted from [25, Chapter 5]. As smooth and efficient flight at low speed is desired, a target AR of 10 is selected as the starting point for the design of the wing geometry. This will be slightly modified throughout the development of the project. Wing planform design For this work, a comparison has been made between 4 possible wing planforms: rectangular, elliptical approximation (Schuemann planform), trapezoidal, and hybrid rectangular and trapezoidal wing. The decision criteria is based on the performance analysis resulting of XFLR5 simulation and building feasibility. The considered wing planforms can be seen on Figure 4.9, and their key parameters are displayed on Table 4.6. The four geometries are further detailed in Annex B. It is important to remind the scale effects mentioned above in Airfoil selection paragraph, 4.2.4, that is why A. Lennon [25] repeatedly advises against the use of chords smaller than 12.7cm (5 inch). (a) Hybrid wing (b) Rectangular wing (c) Schuemann wing (d) Tapered wing Figure 4.9: Compared wing geometries. The top side of the geometries corresponds to the leading edge. Plotted on XFLR5 [28]. 28 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION fuselage, so its result is considered an optimistic approximation. Once this value is obtained, Static Margin is calculated in Table 4.9 for different CG positions. hn=h0+ηH·VH+ (a1/a)·1−dϵ dα (4.14) •hn= position of the neutral point as a decimal fraction of the wing MAC. •h0= position of the aerodynamic center of the wing on the standard wing chord. h0= 0.25. •ηH = stabiliser efficiency. ≃ 0.9 for a T-tail, ≃ 0.6 for a normal tail, ≃ 0.4 or 0.3 if tail is near wing wake or on a fat fuselage in disturbed flow. ηS= 0.6. •VH= tail volume coefficient, estimated in 4.11.VH= 0.6 •a1 = slope of the lift curve of the stabiliser. Computed through Eq. 4.15 [26] for AR = 4, since this is the target AR of the stabilizer. a1= 0.073deg−1. •a = slope of the lift curve of the wing. Computed through Eq. 4.15 [26] for wing’s AR. a = 0 . 93 deg−1 . •dϵ/dα = change in stabiliser downwash angle versus change in wing angle-of-attack, typically 0.5 to 0.33. dϵ/dα = 0.5. acorrected =a∞ 1 + 18.25 AR ·a∞(4.15) XNP [m]XNP as MAC% XCG as MAC%XCG [m] 0.068 0.39 25 0.044 0.025 0.14 30 0.053 0.016 0.09 XNP −XCG [m] Static Margin Table 4.9: Neutral Point resulting from Eq. 4.14. Static Margin is reduced as the CG is pushed aft. Looking at the results, the CG is placed on 30% MAC , providing a Static Margin of 0.09. Healthy Static Margin values go between 0.05 and 0.15, according to references [25, p.36][26, p.174, p.250-252]. Therefore a CG on 25% MAC could be also a suitable option in case of some more stability were required. Now, Tail Moment Arm ( TMA ) can be calculated for balance the mass forces on the desired CG point. This task is performed in Section 4.2.11, Weight centering, by taking into account all the components and structures forming the aircraft, and its distance to the desired balance point. A TMA = 0 . 676 m is obtained. Then, using Eq. 4.12 the Horizontal Tail Surface ( SH ) is calculated, and with it all the geometrical parameters of the tail. Those are displayed on Table 4.10. Finally the incidence angle of attack of the tail is reviewed using XFLR5. For this, the wing-tail assembly is modelled in the simulation environment, trying different angles of attack of the tail until the curve Cm - 35 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION alpha passes through the (0,0) point, thus obtaining the equilibrium in level flight attitude. This procedure can be seen on Figure 4.17. Figure 4.17: How tail incidence affects airplane’s moment coefficient. CG on 30% MAC . Green plot corresponds to stabilizer incidence of -0.2 degrees, red for -2 degrees. Plotted on XFLR5 [28] Airfoil NACA 0008 bH[m] 0.475 VH0.6 croot [m] 0.120 TMA [m] 0.676 ctip [m] 0.100 SH[m2] 0.0522 ARH4.32 α0H[◦] -0.2 Table 4.10 For adding contrast, Neutral Point can also be calculated by XFLR5 as XNP = d ( XCp.Cl ) /dCl . For the defined configuration, XFL5 is giving a XNP = 0 . 095 m , what would offer an incredibly optimistic static margin of 0.21, far after the one estimated through Eq. 4.14. All the values used on Equation 4.14 have been extracted from, or are common to, XFLR5 model, except for ηH factor. Thus, this difference on NP position may be because XFLR5 model is not considering any tail efficiency reduction factor, resulting in a very optimistic estimation. That said, the more conservative Neutral Point value is taken as good. 4.2.7 Vertical tail and lateral stability A well dimensioned vertical plane plays a vital role in the stability of the aircraft, on the yaw and roll axes. A too small vertical stabilizer combined with a large dihedral can result in lateral oscillations, also called “Dutch Roll”; a vertical stabiliser too large compared to the dihedral can lead to spiral instability. As in horizontal stabilizer, the vertical tail has been designed using NACA 0008 airfoil, and built as a flat plate. Two methods are compared to estimate the Vertical Tail Surface ( SV ). The first one sizes the fin surface based on recommended values of the Vertical Tail Volume ( VV ) (Equation 4.17), for the desired flight attitude 36 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION and aircraft configuration. Daniel P. Raymer tabulates on [8] a VV value of 0.02 for gliders and high AR airplanes. It is worth saying that 0.02 is the minimum of all the tabulated VV values for all different aircraft configurations. Once the VV coefficient is selected, the geometry of the vertical tail plane is defined (remember TMA = 0 . 676). An Aspect Ratio from 2.5 to 3, is recommended on [25, Chapter 9], and Equation 4.16 is provided to calculate it. The chords and span are manipulated to meet the desired AR , while keeping chord as high as possible to avoid adverse viscous effects. ARV=1.55 ·bV SV (4.16) VV=SV·(TMA) S·b(4.17) Once a first estimate of the vertical tail is designed, the lateral area rule is applied. This method was introduced by Charles H. Grant in his book Model Aircraft Design, published in 1941. Grant states that the Center of Lateral Area ( CLA ) should lie on a horizontal line through the Center of Gravity of the model in a level flying attitude, and about 30 to 35% of the distance from the CG to the Aerodynamic Center of the vertical tail surface, i.e. the TMA . Thus, the first step is to locate the CLA of the airplane. The vertical tail estimated through Vertical Tail Volume method is taken as starting point; then, using a 1:1 CAD model of the airplane, designed on Solid Works, a side view of the complete model is drawn and extruded on constant thickness (Figure 4.18); the surfaces which are duplicated on the right and left side of the model, like dihedral, landing gear, and wheels, are doubled in thickness. The CLA corresponds to the center of gravity of the solid, which can be easily computed by Solid Works. The first estimated vertical tail places CLA on 34% of the mentioned line (Figure 4.19a). Then the vertical stabilizer is trimmed to move that point into 30% of TMA . Finally a dorsal fin of 10% of SV is added, as indicated in [25, Chapter 9], to improve tail effectiveness and increase attaching surface to the tail boom (Figure 4.19b). The final result is a center of lateral area on 31% TMA and a Vertical Tail Volume of 0.018. It must be acknowledged that this method was initially applied incorrectly, resulting in a very small area of the vertical stabilizer. This error was discovered after Test Flight III 6.3, during which the crash partially occurred due to insufficient rudder authority during takeoff. (a) CAD model of the full airplane. (b) Projected lateral view of the model for area rule application. Dihedral, landing gear and wheels he been doubled thickness. Figure 4.18: CAD model drawn on Solid Works. 1:1 scale. 37 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION (a) First estimated vertical tail plane. VV = 0 . 2. CLA on 34% of TMA. (b) Previous tail is trimmed to bring CLA forward up to 30% of TMA . Then the dorsal fin added leading to the definitive vertical tail surface: VV = 0 . 018, CLA on 31% of TMA Figure 4.19: Area rule application to determine the vertical tail surface and geometry. The three vertical lines correspond to 25, 30 and 35% of the TMA. Airfoil NACA 0008 Dorsal fin SV[m2] 0.0152 SF0.0015 bV[m] 0.16 Base [m] 0.07 croot [m] 0.11 Height [m] 0.043 ctip [m] 0.08 AR 2.611 VV0.018 Table 4.11: Final design parameters for vertical tail plane. 4.2.8 Control Surfaces Manoeuvring authority for the aircraft is achieved through the control surfaces, which are located at the trailing edge of the wing, the horizontal stabiliser and the vertical tail stabiliser; being those the ailerons, elevator and rudder. These surfaces must be sized to provide the desired amount of control action to ensure adequate manoeuvrability. This sizing can be done using flight mechanics equations, but a much simpler method is chosen. This tabulates common values of the dimensions of the control surfaces as a function of the geometrical parameters of the aerodynamic surface containing them. These parameters are found in the Lennon reference book, and are shown in the Table 4.12. Parameter Elevator Rudder Ailerons (one on each wing tip) Main Surface % 30 50 - Area [m2] 0.0157 0.0076 - 0.0150 c [m] 0.041 0.048 0.16 c 0.026 b [m] 0.382 0.16 0.6 b/2 0.584 Max. Deflection [◦]±25 ±30 +25, −10 Table 4.12: Design values for the control surfaces, following. Cells filled with “-” correspond to non-interest values. 38 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION 4.2.9 Landing gear Initially it was not intended to add landing gear to the aircraft, as it was conceived to be hand-launched and soft-landed. This is why landing gear weight has not been considered on initial weight estimation on Annex A. When the design was already advanced, it was decided to add landing gear to protect the aircraft from hard landings on the first flights, where the aircraft is not yet calibrated. Finally the landing gear has ended up being used for both take-off and landing. The tail sitter configuration has been chosen, as it is the simplest, replacing the tail wheel with a simple wire that avoids contact between the stabiliser and the ground. This saves weight and complexity for the tail, but means that yaw cannot be controlled during take-off, until sufficient speed is reached for the vertical fin to gain authority. For the positioning of the landing gear, the guidelines provided on [25, Chapter 16] have been followed, mounting it as shown in the following Figure: Figure 4.20: Constructive parameters on the landing gear assembly. Extracted from [25, p. 75]. 4.2.10 Power plant and energy management As it is explained in Section 4.1.3, the only propulsion plant conceived for this project is the electric one, due to its simplicity and reliability. Therefore the propulsion plant of the developed aircraft consists of the battery module, the Electronic Speed Controller ( ESC ), and the engine, which is direct driving the propeller. It is common some UAV s to mount separated battery modules for adding redundancy to the electric power supply. It is also usual to use separate battery modules to power the propulsion plant and on-board electronics for the same reason. In this case, one battery module is feeding both the propulsion plant and the on-board electronics. Available batteries at the time of developing the project are LiPo 3S, 11.1V, 1500mAh manufactured by Ovonic, which are stacked in parallel to increase capacity up to 3000mAh, while adding some level of redundancy against a direct battery failure. The appropriate procedure to determine the energy requirements at an early stage of the design would be through the estimation of the drag coefficient of the whole aircraft. Then the aerodynamic drag in cruise flight would be calculated, leading to an estimation of the power needs of the aircraft and the calculation of the required battery capacity to meet the desired flight endurance. As discussed in Section 4.1.3, available tools would not give a realistic estimation of drag coefficient. Therefore, conclusions about energy performance are made after analyzing the data collected on Test Flights, Chapter 6. However, it is still necessary to select a suitable engine for this first flight test phase. This choice is based on typical values of the thrust-to-weight ratio (T/W) of existing model aircraft, which provides information on 39 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION the required thrust that an aircraft with similar flight characteristics should have. Thus, common T/W ratio for gliders and trainer airplanes are set between 0.35 and 0.55. As the final airplane’s mass is 1.62kg (Table ??), required thrust is around: T=m·g·T/W = 1.62 ·9.81 ·0.55 = 8.74N(4.18) That said, the motor among the available ones that better matches the required thrust is Flash Hobby D3536-8 1000KV (Figure 4.21), with a nominal maximum pull of 11.38N what represents an increase of 130% over what would be theoretically required. This sets the actual thrust-to-weight ratio on T/W = 0.72. The motor is installed 2 ◦ right and 2 ◦ down relative to the fuselage axis to overcome the propeller’s inertia and the nose-up effect of the propeller’s wake. Figure 4.21: Manufacturer specifications for D3536-8 1000KV electric motor. In the Figure above it can be seen that an engine with the same dimensions has several sub-models with different KV values. The KV ratio of an electric motor corresponds to the RPM (revolutions per minute) of the motor per volt applied. Therefore a motor with a higher KV will rotate at a higher speed when a certain voltage is applied. However, motors with higher KV tend to generate less torque. This is a trade-off between speed and torque. In Figure 4.21 can also be seen that specific propeller is recommended for each motor KV. Propeller geometry is usually defined by two simple parameters: propeller diameter ( D ) and advance pitch ( p ), which are commonly provided as D×p . Advance pitch is the theoretical axial displacement in one turn as if a reference blade-section was screwed on a solid. Those parameters are often given in inches. The closest option available to the recommended propeller is a 11x6 geometry, what will provide slightly more thrust for the same RPM and wind speed than the recommended 5 inch-pitch version. Motor specifications also recommend anElectronic Speed Controller ( ESC ) capable of delivering up to 50A. As the motor will be running on 11.1V, the maximum current to be delivered to the motor when operating at maximum power (410W) will be 36.94A, as P = V·I . Taking into account an esc efficiency of 95%, a 40 4.2. PRELIMINARY DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION maximum current value of 38.88A is obtained. Thus, the used ESC is Tekko 32 F4 45A, which allows for a safety margin of 15.7% over the maximum theoretical demand. An important feature is its capability to output telemetry, providing the flight controller with a readout of the power consumed and the angular velocity of the motor, which can be later used on Test Flights to calculate the actual performance of the aircraft. 4.2.11 Weight centering Weight centering, also known as Center of Gravity ( CG ) management, holds significant importance in the design and functioning of aircraft. It involves ensuring that the aircraft’s center of gravity remains within specified limits, ensuring flight stability and manoeuvrability. Thus, an accurate distribution of the weights in design phase will avoid the need of extra weights to balance the CG once the aircraft is built, thus preserving the efficiency of the aircraft. For this task, the weight of all components and structures of the aircraft has been noted, together with the distance from the center of mass of the component to the desired center of mass of the assembly, defined to be on 30% MAC on Section 4.2.6. A 3D model of the complete aircraft, which can be seen on Section 4.3, is used to distribute all the electronic components and power plant inside the fuselage. Then, by trial and error, the components and structures are adjusted to achieve a resultant moment of 0 at the CG of the aircraft. The parts with greatest effect on this balance are the tail on the back side, and the engine and batteries on the front side. The Tail Moment Arm is set first to roughly compensate for the moment produced by the front of the aircraft. Then the engine and batteries are adjusted to achieve balance at the desired point. With this, Annex C shows the list of all components that make up the UAV , together with their mass and their position relative to the specified Center of Gravity. With the final value of the aircraft mass provided by the table in Annex C, the main aircraft parameters are updated to their final values. The following Table therefore contains the parameters which essentially define the performance of the UAV of this work. 41 4.3. DETAILED DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION Operational Specifications Total Mass [kg] 1.652 Wing Loading [kg/m2] 4.92 Wing Area [m2] 0.3357 Wing Span [m] 1.95 Aspect Ratio 11.3 Cruise Speed [m/s] 14.58 CG in MAC% 30 Endurance [min] 14.2 Airframe Dimensions Length [m] 0.983 Width [m] 1.95 Height [m] 0.188 Table 4.13: Main parameters of the final aircraft, once built. 4.3 Detailed design Since the entire design and construction process of this aircraft has been approached as an iterative process, the preliminary design, detailed design, and construction have proceeded almost concurrently. This is exemplified, for instance, in the second iteration regarding the wing area (Table 4.7), which was based on the weight of the central wing section already built at this point. Portability has been a priority from the beginning of the initial concept, leading to the division of the aircraft into distinct detachable modules: fuselage, wing, divided into three sections, and tail. 3D printing has played a crucial role in connecting these modules, as well as accommodating the electronics and motor within the fuselage. These components have been designed from scratch to fulfill their intended purposes. To streamline the design and construction process, the entire aircraft has been assembled within the Solid Works environment, which has proven highly beneficial for designing auxiliary parts and ensuring proper weight distribution. To ensure rigidity between the modules, carbon fiber tubes and rods are used in the joints. 4.3.1 Fuselage module As already explained in Section 4.2.3, the fuselage is the initial component to be designed, with minor modifications throughout the design process. Consequently, the final version of it is presented here, along with the components and attachments that secure the motor, landing gear, electronics, wing, and tail. The wing is attached to the fuselage using heavy-duty rubber bands tied to two 4mm carbon fiber tubes, mounted in front and behind of the wind. The final fuselage module can be seen on Figure 4.22c. 42 4.3. DETAILED DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION (a) Final version of the fuselage main core. Made out of foam board. (b) 3D printed parts inside the fuselage. From left to right: motor mount, landing gear mount, Raspberry Pi mount, GPS and radio antennas mount (right top), and tail mount. (c) Complete fuselage module. The main body is set to translucent to show the interior. Inside fuselage can be accessed through the front cover (fixed with tape) and rear cover (fixed with one M3 bolt), as well as through te wing opening. Figure 4.22: 3D views of the different fuselage parts. (a) Motor mount. Lateral view (b) Motor mount. Top view. Figure 4.23: Detail of motor mounting incidence: 2◦down, 2◦right. 4.3.2 Wing module As the wing span is 1.95m, it is divided into detachable modules to facilitate transportation (Figure 4.24a). The side sections each have two 6mm carbon rods: one at the thickest point of the airfoil, which absorbs all joint loads, and a shorter one at the rear, which acts as a torsion pin. The central section is closed at the sides with a 3D printed part, which helps with the placing of the attachment points when building the wing section. Both connecting pins of the side sections fit into two 8x6mm diameter tubes in the central module, Figure 4.24c. The lateral displacement of the sections is fixed with a strip of adhesive tape. When the wing is assembled, the central part of the structure is reinforced by a carbon fibre spar that covers 50% of the 43 4.3. DETAILED DESIGN CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION total span. As it has been already said in Section 4.2.5, dihedral has not been considered in the initial design phase, due to limitations in the available building methods. However, a 5 degree dihedral is still effective due to the bending of the wing outer sections under the action of the weight of the aircraft. This bending effect can be appreciated in Figure 4.35. (a) The wing is formed by three detachable modules. (b) Internal structure of the wing central module. (c) Detail on wing modules attachment. Figure 4.24: Different views on the three wing detachable modules. 4.3.3 Tail module The tail module consists of the tail boom (13mm diameter and 1mm thickness carbon fiber tube), the attachment to the fuselage, the tail surfaces, reinforcements, and the tail skid. The attachment to the fuselage consists of a 3D-printed piece inserted into the tail boom and glued with epoxy; depicted in Figure 4.26. The tail surfaces are made of foam board, and the reinforcements are 3D-printed components joined with hot glue. The tail skid is a 1.5mm stainless steal thread. (a) Full tail module. (b) Detail on tail reinforcements and tail skid. Figure 4.25: Tail module views. 44 4.4. BUILDING PROCESS CHAPTER 4. AIRPLANE DESIGN AND CONSTRUCTION 4.4.4 Conclusion of the building process The construction process has resulted in a dismountable yet robust aircraft, capable of carrying out the missions defined in Chapter 6and surviving three forced landings. Nevertheless, certain aspects of the design and construction that did not turn out to be optimal should be highlighted. The decision to divide the wing into three segments has proven to be inappropriate. With an initial estimated mass of 0.401 kg, the mass of the constructed wing is 0.594 kg, representing a deviation of 48%, which is unacceptable. This is primarily due to the presence of carbon tubes required to secure the two junctions, as well as the adhesive and 3D prints that hold them in place. Furthermore, the construction nature of these three segments inevitably leads to the observed flexion as depicted in Figure 4.35. Dividing the wing into two segments would reduce the number of junction elements and, consequently, weight, also leading to a more robust structure since the wing’s attachment to the fuselage would simultaneously secure both wing sections. A reduction in weight translates to a reduction in wing loading, making hand launch a potential option. This would eliminate the need for both landing gear and associated fuselage reinforcements, resulting in significant weight and aerodynamic resistance savings for the aircraft, while also reducing construction complexity. Another factor adversely affecting the vehicle’s efficiency is the presence of duct tape on the wings. Its rough nature is counterproductive in maintaining laminar boundary layer flow during flight, thereby increasing drag. Therefore, for future iterations, it is proposed to use thinner tape with a smooth surface finish and to minimize its usage. 51 T´ıtol del TFE Chapter 5 Electronic systems As introduced at the beginning of this document, the mission of the developed aircraft is to orbit autonomously over a second UAS in order to relay its signal and allow its operation in Beyond Visual Line Of Sight ( BVLOS ) missions. For this purpose, the designed is equipped with two separated processing units: the flight controller and the digital link unit. The first one consists of a Pixhawk flight controller board, conceived for autonomous UAV navigation, that uses internal and peripherial sensors to process the aircraft attitude and send an output response to the actuators. The digital link unit consists of a Raspberry Pi board equipped with an HD camera and a transmission (WiFi) module. Raspberry Pi is connected via UART serial port to the flight controller and is responsible for the communication of the UAS with the ground control station in a bidirectional link, sending HD video and flight controller’s telemetry to the ground and receiving control commands from the pilot. Therefore, this chapter develops the operation of all the electronics with which the aircraft is equipped, as well as their specific function and the decision-making associated with their presence on-board. 5.1 Basic electronics: propulsion plant and actuators In traditional Radio Controlled ( RC ) models, the aircraft is piloted from the ground via a radio transmitter, which encodes the pilot’s commands and sends them to the on-board receiver. This receiver decodes this signal and outputs the raw orders to the actuators via PWM signal. The receiver has several PWM output channels where the actuators are connected. The most basic electronic configuration for an electric RC airplane consists of a battery pack, a Battery Eliminator Circuit ( BEC ), a motor, an ESC , the servomotors, and a radio receiver. The BEC downs the battery voltage to 5V to feed the receiver and the servos, and the ESC is in charge of controlling the motor’s power supply. BEC and ESC often are combined into the same electronic component. This basic wiring has been installed in the aircraft on its first test flight, with the pilot taking over a fully manual flight. The diagram representing it can be found in the Annex E. 52 5.2. AUTONOMOUS FLIGHT CONTROLLER (FC) CHAPTER 5. ELECTRONIC SYSTEMS 5.2 Autonomous Flight Controller (FC) A flight controller board, often simply referred to as a Flight Controller ( FC ), constitutes the central processing unit of the aircraft, responsible for stabilizing and controlling the vehicle’s flight. As the UAS industry is an on growing sector, many of the main drone manufacturers are producing their own flight controllers, equipped with different functionalities and customization levels to fulfill their mission. However, there is a list of fundamental attributes common among all flight controller boards: • Stabilization: Boards include gyroscope and accelerometer sensors to observe and manage the attitude of the UAV . The level of attitude control can be selected by the pilot through the different flight modes. • Global Navigation Satellite System ( GNNS ) integration: It is common to incorporate a GNNS antenna module, enabling features like waypoint navigation, return-to-home functionality, and geofencing. GNNS data is used by the attitude control unit for accurate position, altitude, and ground speed information. • Communication: Radio modules are used to establish wireless communication with Ground Control Station ( GCS ). This allows pilots to send commands, receive telemetry data, and monitor the UAV ’s status during flight. • Autonomous capabilities: Advanced flight controllers can execute pre-programmed flight paths, follow specific GPS coordinates, or perform complex maneuvers autonomously. • Safety features: all the above listed features enable the aircraft to autonomously respond to dangerous situations, such as power loss, low battery, or telemetry loss; ensuring the safety of the aircraft and the environment. • Customization: Flight controller boards allow for implementation and communication with other onboard computers, capable of processing images and complex algorithms. Those can enable functionalities such advanced autonomous flight functionalities or on-board AI -powered computer vision, among others. A quite popular choice for the flight controller board in autonomous UAV applications is PixHawk. Indeed, most of the academic papers focused on advancing technology and algorithms for autonomous UAV s rely on this platform [30]. This board has all the capabilities described above, but there are three major reasons for this choice: (i) PixHawk board was one of the pioneering open-source platforms for drone technology, and today remains as one of the market standards; (ii) has robust control algorithms, for multiple vehicle types, and the ability to reach deep levels of customization; (iii) features i2C, UART, CAN and SPI ports for enabling serial communication with external sensors and on-board computers, allowing for multi-level architecture; (iv) has a very extensive documentation and accounts for a huge community support in both the hobby and professional sectors. The software that powers this board is ArduPilot, born in 2009, now established as one of the most respected open-source autopilot platforms in the industry. Ardupilot [1] provides a complete and flexible solution for the autonomous control of multiple platforms, including multicopters, aircraft, rovers or submarines; and accounts for an extensive and detailed documentation explaining every of its features and customization levels, as well as a full compatible hardware review. 53 5.3. SENSORS CHAPTER 5. ELECTRONIC SYSTEMS 5.2.1 PixHawk board Pixhawk is a generic term that refers to a family of open-source autopilot hardware platforms. These platforms are designed to run autopilot software like Ardupilot or PX4 (a closed licence autopilot software based on Ardupilot). There are different versions and manufacturers of Pixhawk hardware, where the products that make up the current standard can be found on the official Pixhawk project website. The specific board used in this project is Pixhawk 1, manufactured and sold by 3D Robotics (3DR), and all its specifications, ports definition and connector guide can be found in Pixhawk 1 description site, on Pixhawk project’s website [24]. In the Ardupilot Complete Parameter List portal [31], a comprehensive list of the configurable parameters on the flight controller is provided along with relevant information, providing control over the extensive levels of customization offered by the system. The parameters that have been configured for the operation of this project’s aircraft are detailed in Annex D. Figure 5.1: Pixhawk 1 flight controller board, together with description of its ports and connectors. Extracted form [24]. 5.3 Sensors Flight controller is equipped with internal and external sensors that provide real-time information about the drone’s orientation, position, speed, environmental conditions, and status parameters, allowing the processing unit to maintain stability and execute flight commands. For enhancing safety and reliability, some of the critical sensors, as well as power supply, are installed redundantly. That said, the sensors installed in the current configuration are listed below, together with the module or board where mounted and the model or identifier. 54 5.3. SENSORS CHAPTER 5. ELECTRONIC SYSTEMS Sensor Module Component Specification Gyroscope Pixhawk 1 ST Micro L3GD20H 16 bit Accelerometer/magnetometer Pixhawk 1 ST Micro LSM303D 14 bit Accelerometer/gyroscope Pixhawk 1 Invensense MPU 6000 3-axis Barometer Pixhawk 1 MEAS MS5611 GNSS GPS module U-blox NEO-7 module Magnetometer GPS module HMC5883L digital compass Battery voltage sensor Power module PM02 Power Module Battery current sensor Power module PM02 Power Module Airspeed sensor + Pitot tube Arspd sens. module Arspd-4525DO ESC telemetry ESC Tekko32 F4 45A ESC HD video Digital link system OV5647 camera module Table 5.1: Sensors fitted to the aircraft, together with the module to which they belong and their technical specification. Gyroscope, accelerometer, and magnetometer provide essential data about the aircraft’s orientation and motion. By combining information from these sensors, the flight controller can precisely control the aircraft’s attitude, altitude, and heading. Due to their critical role, these three sensors are the only ones with redundancy in the basic autopilot configuration. Onboard barometer provides altitude data based on atmospheric pressure. GNNS antenna provides information about the position, ground speed and altitude to the autopilot. It enables features as autonomous way-point missions, terrain obstacle avoidance and return to home capability. GNNS data merges with gyroscope, accelerometer, magnetometer and barometer to provide more accurate position, altitude and velocity measures. The used GNNS module also includes a magnetometer to provide aftermentioned redundancy. The airspeed sensor measures the upsetream flow velocity by comparing static and dynamic air pressure. This enables throttle to control the target airspeed in the defined range (maximum and minimum airspeed) in autonomous flight modes. Power consumption is measured by the Power Module and the ESC . Power module is a mandatory hardware on Pixhawk board setup, and provides readout of the power output (voltage and current) of the battery. In the other hand, the mounted ESC features BLHeli32 Dshot 1200 firmware that enables telemetry output. ESC telemetry provides motor power and RPM data, what is then used along with efficiency estimation of the propeller for calculating airplane’s thrust force. Also ESC and Power Module power readings can be compared to calculate the energy consumption of the whole airplane’s electronics. Finally, the system is equipped with an HD camera operated by the digital communication system, explained below in Section 5.4. Although the autopilot is not interacting with this data, live video feed is highly appreciated in Ground Control Station to provide situational awareness to the pilot on long range missions. 55 5.4. RUBY FPV: DATA TRANSMISSION SYSTEM CHAPTER 5. ELECTRONIC SYSTEMS 5.4 Ruby FPV: data transmission system In the case of an UAS with a camera, there are three main data streams between it and its Ground Control Station: the control link, the telemetry link, and the HD video link. In the industrial environment, such as DJI’s Mavic models, the transmission is unified into a single system that bundles the data streams, encrypts them, and sends them through a multi-band system, capable of automatically hopping between the 2.4GHz and 5.8GHz frequencies. This provides more efficient data management, minimises interference and thus increases the operating range. An open-source version of this solution is also proposed by the Ruby FPV system, developed by Petru Soroaga, that was first released on April 25, 2021. Ruby FPV is defined as “a complete platform (hardware and software) designed for controlling and managing UAVs, [...] and offers robust end to end digital radio links between multiple points” [9]. All information pertaining to Ruby FPV can be accessed on its official website [9], where comprehensive details regarding compatible hardware, wiring, assembly, and configuration are covered. The software is provided in the form of a plug-and-play operating system, which is installed on the Raspberry Pi computer through the utilization of a Micro SD card. The user interface of Ruby FPV comprises a screen connected via HDMI to the ground controller, which can be accessed through four push buttons that enable the user to navigate through all its menus and options. Additionally, extra buttons and switches can be installed to enable auxiliary functions. To avoid the use of a second screen, the Ruby controller is connected to the laptop’s display through an HDMI to USB video decoder, which projects the video input into the camera application as if it were a webcam feed. That said, Ruby FPV platform is here used to manage the communications of this project’s UAV , in particular by assuming the following functions: • Telemetry bridge over the Pixhawk Flight Controller and the Ground Control Station, running on Mission Planner. •Live HD video transmission from the UAV to the Ground Control Station. • Remote control link between the pilot and the UAV through the Ruby FPV Ground Control Station interface. •Relay system between the Ground Control Station and the relayed UAS. Figure 5.2. •Data stream monitoring and control over the radio settings through user interface. 56 5.4. RUBY FPV: DATA TRANSMISSION SYSTEM CHAPTER 5. ELECTRONIC SYSTEMS Figure 5.2: Steps to configure relay radio link. Extracted from Ruby FPV website [9] This fully integrated relaying feature within the system is one of the main reasons why Ruby FPV is chosen. With this capability, the developed UAV is able to act as a node between the ground station and a second vehicle, in order to extend its range and enabling BVLOS operation in search and rescue missions. To implement this, only a Ruby control station connected to both vehicles is required, along with an additional radio module in the relayer vehicle to establish parallel and interference-free communication with the ground station and the extended-range vehicle. This feature is not implemented in this work due to a lack of budget to acquire the additional components. Therefore, it remains pending for future phases of the project. As an open-source project, relevant information for customizing both the software and hardware can be found on the Development tab of its website. In this context, Software Development Kits (SDKs) in the C/C++ programming languages are provided to facilitate this customization process. In addition, there is the possibility to contact the team for the development of on-demand functionalities, both for enthusiasts and professionals. Although delving into this aspect is above the scope of this work, the customization possibilities are of interest when considering the project’s continuity and scalability beyond the completion of this work. 5.4.1 Ruby FPV hardware setup The hardware required by Ruby is both cost-effective and readily available. For the vehicle component, it only necessitates a Raspberry Pi computer, a network card or WiFi module with its respective antennas, and a camera compatible with Raspberry Pi, all powered by a 5V source. The ground station, on the other hand, requires a Raspberry Pi computer, a WiFi module, and five push buttons for navigating the user interface. Additionally, a serial to USB converter is included for telemetry communication with a computer, an HDMI to USB video capture device is utilized to display the user interface on the computer, and a BEC is employed to power the circuitry via a 3 or 4s LiPo battery. Regarding the Raspberry Pi computer models, it is recommended to use a compact model such as the Zero, Zero W, or Zero 2W for the vehicle, while a full-sized model is recommended for the ground station to ensure 57 5.4. RUBY FPV: DATA TRANSMISSION SYSTEM CHAPTER 5. ELECTRONIC SYSTEMS maximum decoding speed and, consequently, minimal latency. At the time of purchasing the computers to commence the assembly, there was a shortage of Raspberry Pi models stemming from the COVID-19 pandemic. Acquiring any Zero-line, as well as the 3B+ and 4 models, was difficult and very expensive. The boards with higher processing capabilities that could be obtained at an affordable price were two Raspberry Pi 3B models, which are utilized in both the ground station and the vehicle. That said, the components used on the vehicle and ground controller Ruby FPV modules are listed in Table 5.2, while their wiring diagrams can be found on ´ Annex E. A photo of the built Ruby FPV Ground Controller can be seen below. Figure 5.3: Photo of Ruby FPV ground controller hardware layout. HDMI cable is not connected to the Raspberry Pi for aesthetic reasons. RC controller USB cable is neither connected. Component Units Specification Vehicle Network card. 5.8GHz 1 RTL8812AU Linear antenna 2 5.8GHz 3dB 90mm Raspberry Pi 1 3 B Camera module 1 OV5647, 130 degrees lens Capacitor 1 10V 100µF Controller Network card. 5.8GHz 1 RTL8812AU Linear antenna 2 DJI 5.8GHz 3dB 40mm Raspberry Pi 1 3 B Push button 5 Self built button pad BEC 1 PDB XT60 W/ BEC 5V & 12V Serial to USB converter 1 FT232RL FTDI USB HDMI to USB 1 HDMI video capture Capacitor 1 35V 330µF Table 5.2: Ruby FPV hardware list and specifications. 58 5.5. VEHICLE ELECTRONICS LAYOUT CHAPTER 5. ELECTRONIC SYSTEMS 5.5 Vehicle electronics layout Table 5.3 lists all electronic components included in the air vehicle, together with their function/group to which they belong and their technical specification. The wiring diagram of the vehicle electronics can be found in Annex E. A picture of the components installed inside the aircraft fuselage is also shown below. Figure 5.4: Photo of the electronic components of the vehicle mounted in the fuselage. Component Units Group Specification Bttery 2 Propulsion / Power 3s 1500mah 100c Motor 1 Propulsion D3536 1000KV ESC 1 Propulsion Tekko 32 F4 45A ESC BEC 1 Power Assan UBEC 8A Servo 4 Actuators MG90S Power module 1 Power / FC / Sensor PM02 Power Module Pixhawk FC 1 FC Pixhawk 1 GPS + compass 1 FC / Sensor 3DR GPS + compass module I2C splitter 1 FC I2C splitter Air speed sensor 1 FC / Sensor arspd-4525DO Raspberry Pi 1 Ruby FPV 3B 5.8GHz WiFi module 1 Ruby FPV RTL8812AU HD camera 1 Ruby FPV / Sensor OV5647, 130 degrees lens Linear antenna 2 Ruby FPV 5.8GHz 3dB 90mm Capacitor 1 1 General 35V 1000µF Capacitor 2 2 General 10V 100µF Zener diode 1 General 5.6V 1N5339 Table 5.3: Electronic components fitted to the airborne vehicle, together with their classification according to the function/group to which they belong and their technical specification. 59 5.6. GROUND CONTROL STATION CHAPTER 5. ELECTRONIC SYSTEMS 5.6 Ground Control Station The Ground Control Station ( GCS ) consists of the ground part of the UAS system that sends the control orders the aerial vehicle, receives its telemetry and monitors its status. Although multiple Ardupilot-based GCS software options exist, at the time of writing only three of these options have recent updates, being Mission Planner, QGroundControl, and MAV Proxy. Those are full featured with Ardupilot capabilities, but they differ in their scope of use. Mission Planner has an accessible and comprehensive interface and is the most widely used by beginners and intermediate users. It runs on Windows only. QGroundControl offers a simplified and more user-friendly interface, and stands out for its availability for Android platforms, being also available on Windows, macOS and Linux. Finally MAVProxy is a command-line GCS that is highly customisable, being particularly useful for advanced users who prefer scripting and automation. MAVProxy can run on Windows, macOS, and Linux. This project will be using two GCS systems for unlocking a wider range of capabilities: Mission Planner and Ruby FPV. Mission Planner will be used as the main GCS for mission and vehicle monitoring, running on the computer. Ruby FPV (further discussed into Section 5.4) will be used, in parallel, as radio platform for data communication and data flow monitoring between the UAV , Mission Planner and the pilot. That said, Ruby FPV Controller constitutes the hardware of the Ground Control Station, apart from the computer; and Ruby FPV software is used to operate all issues related to radio link and data transfer. 5.6.1 Mission Planner GCS software Mission Planner is a widely used GCS software for Ardupilot. It’s open-source and runs on Windows. Mission Planner allows for: • Mission planning: On Mission Planner’s Plan window, waypoints, flight paths, altitude, and other parameters can be set to define the mission objectives. • Pre-flight checks: Before launching the vehicle, the Data tab offers pre-flight check by providing a customisable checklist to the operator. Some of this checklist’s parameters include sensors health, automatically monitored by the GCS. • Real-time flight monitoring: During the mission, Mission Planner Data tab allows for UAV ’s status monitoring through vehicle’s telemetry data, that includes attitude, speed, GPS coordinates, battery status, and camera feeds, among others. • Safety and emergency response: In the event of emergencies or unexpected situations, the GCS can trigger safety protocols, such as return to home or emergency landing. • Post-mission Analysis: Once the UAV finishes its mission and returns to the ground, Mission Planner provides an interface designed to download and analyse flight data stored in the flight controller’s memory, by providing detailed flight logs, sensor data, and mission reports. This functionality is utilized in Section – to scrutinize test flights and formulate conclusions regarding the aircraft’s performance. 60 6.5. FLIGHT V CHAPTER 6. FLIGHT TESTS only focused on the parameters and position displayed in Mission Planner and the FPV camera of the Ruby system. The landing was performed with FBWA assistance activated, resulting in a smooth and precise touchdown. Relevant data associated at the takeoff and landing behaviour of the aircraft is displayed in the table below. Take off Landing Flight mode Manual FBWA Used runway [m] 15 120 Wind speed [m/s] 12.8 11.5 Flapperon deflection [%] 25 0 Table 6.1: Take off and landing relevant data obtained from flight log. Distances have been measured with the help of Google Earth ruler tool. 6.5 Flight V - Place: Club RC Plana Baixa, Cabanes - Flight modes: MANUAL, FBWA, FBWB - Date: September 17, 2023 - Flight time: 9min 36s - Rating of the flight: Successful - Power consumed: 19.44 Wh This flight is used to collect more information about the performance of the aircraft. Here, different level flight passes are carried out using the FBWB flight mode. Figure 6.5: Flitht path of Test Flight V. Waypoints are displayed as the previous flight’s mission has not been erased from Pixhawk before the flight. 67 6.6. FLIGHT TESTS CONCLUSIONS CHAPTER 6. FLIGHT TESTS 6.6 Flight tests conclusions The data collected during Flights IV and V were intended for empirical calculations of the aircraft’s performance during flight. Nevertheless, after processing the obtained data, the need for a robust theoretical foundation and experimental methodology has become evident in order to conduct the study properly. Furthermore, it becomes apparent that a significantly larger amount of data, than what has been collected in these two flights, would be required to carry out this analysis adequately . Therefore, before reaching misleading conclusions, it is declared as pending for future studies to rigorously perform this analysis, since the possibility of determine the aircraft’s performance in real-flight situations is of great value in optimizing the design, while also enhancing the learning process derived from it. Nonetheless, the data on total energy consumption and flight time from Flights II, IV, and V are used to calculate the maximum flight time of the aircraft. By calculating the flight time per unit of energy consumed for each case and taking the average, a value of 28.42s/Wh is obtained. Knowing that the batteries used provide 33.3Wh, and reserving a safety margin of 10% of the energy for emergency situations, a maximum flight time of 14 minutes and 12 seconds is obtained. This falls short of the 30 minutes defined as the objective in Section 4.1.1. The next step, therefore, would be to experimentally calculate the coefficient of aerodynamic resistance of the aircraft, analyze its improvement margin, and based on this, determine the battery capacity required to meet the 30-minute flight requirement, stated in Top Level Requirements, Section 4.1.1. It should also be noted that the use of the UAV as a replay node using Ruby FPV in the flight tests is not mentioned, although in the initial parts of the work it is stated that this is its purpose. This is because this system, although proposed, has not materialised due to a lack of financial resources. With that said, the success of this part of the work relies on the proper functioning of the systems implemented in the aircraft, ensuring that the specifically designed vehicle is capable of integrating the Ardupilot and Ruby FPV systems. It is the combination of these three main components, the vehicle, the autopilot, and the radio link, what makes the successful execution of an autonomous mission possible. 68 T´ıtol del TFE Chapter 7 Budget summary One of the objectives of this work has been to maintain a tightly controlled budget associated with the development of the UAV . The majority of the materials and electronics costs have been borne by the student personally, including materials for the vehicle’s construction, electronics, and consumables, adding 530.12 e in total. The Pixhawk control board and some of its sensors were provided by the LUAS laboratory at ESEIAAT. Additionally, thanks are extended to the Trencal`os Team for providing access to their workshop, along with the tools and personal protective equipment (PPE) it contains. The costs derived from the use of the workshop and the tools required are found taking into account the annual depreciation of the material and the monthly cost of the workshop. Regarding the hours of work invested, an estimated total of approximately 620 hours has been allocated to research, design, construction, electronic assembly, software configuration, flight testing, and report writing. The labour rate used for costing purposes is set at 8 e /h, as this is the hourly wage recommended by the university for paid work. Thus, a summary of the project costs is presented in the table below. Item Cost [e] Work hours 4960 Material and electronics 701 Workshop and tools 840 TOTAL 6501 Table 7.1: Summary table of the total economic costs of the project. 69 T´ıtol del TFE Chapter 8 Conclusions In conclusion, the success of the design, construction, and autopilot integration of the developed UAV is demonstrated by its successful completion of the flight testing phase. Thus, this project has encompassed the integral realisation process of the first prototype of an aircraft capable of performing autonomous missions, by integrating open source software and hardware with an aerial vehicle made with cheap materials and simple techniques. The implementation of the Ruby FPV platform alongside the Pixhawk autopilot provides the capability to monitor and control the vehicle’s communications with the ground station in an integrated manner, consolidating telemetry, digital video, and control link into a single data link. Moreover, by utilizing Ruby FPV’s ability to simultaneously control multiple aircraft as relay nodes, the developed aircraft has the potential to act as a relayer between a ground station and a second vehicle, enabling Beyond Visual Line Of Sight (BVLOS) operations, with clear applications in challenging terrain search and rescue missions. Regarding the constructed vehicle, it remains a prototype that requires optimization before it can be considered competitive. In the conclusions of the design and construction process, it is noted that some decisions made on paper have resulted in increased weight or aerodynamic resistance in the aircraft, reducing its flight time. An example of this is the division of the wing into three sections, resulting in a 48% higher weight than initially estimated, which is not acceptable. However, the result of this construction has ended up being both robust and detachable, and has survived two crash landings. So, despite the aspects that could be improved in the construction and design, the result of the construction is considered satisfactory for a first prototype. Furthermore, the development of a rigorous and well-founded method for experimentally analyzing the aircraft’s performance is still pending. In other words, the ability to accurately calculate the various key performance parameters of the aircraft through data provided by onboard sensors, under different flight conditions, would provide a clear and comprehensive picture of how far the design is from being optimal, as well as an idea of the maximum performance that the employed low-cost materials can offer. Overall, the work carried out over the months of this project has proven to be of great educational value, applying the engineering process to develop a prototype solution with clear real-world applications. Thus, 70 CHAPTER 8. CONCLUSIONS further development of the prototype presented here opens doors to potential collaborations and commercialization opportunities within a continuously growing market where UAV s find interesting applications in sectors such as search and rescue operations, aerial surveying, environmental monitoring, precision agriculture, and beyond. 71 T´ıtol del TFE References 1. TEAM, Ardupilot Development. Autopilot Documentation. 2023. Available also from: https : / / ardupilot.org/ardupilot/index.html. 2. SAEED, Adnan S.; YOUNES, Ahmad Bani; CAI, Chenxiao; CAI, Guowei. A survey of hybrid Unmanned Aerial Vehicles. Progress in Aerospace Sciences [online]. 2018, vol. 98, pp. 91–105 [visited on 2023-06-08]. issn 03760421. Available from doi:10.1016/j.paerosci.2018.03.007. 3. COHN, Pamela; GREEN, Alastair; LANGSTAFF, Meredith; ROLLER, Melanie. Commercial drones are here: The future of unmanned aerial systems [online]. 2018 [visited on 2023-06-08]. 4. JOUAV. Oil Gas [online]. [N.d.]. [visited on 2023-06-08]. Available from: https://www.jouav.com/ industry/oil-and-gas. 5. SONKAR, Sarvesh; KUMAR, Prashant; PULI, Yuvaraj Tanjore; GEORGE, Riya Catherine; PHILIP, Deepu; GHOSH, Ajoy Kanti. Design & Implementation of an Electric Fixed-wing Hybrid VTOL UAV for Asset Monitoring. Journal of Aerospace Technology and Management [online]. 2023, vol. 15, e0823 [visited on 2023-06-02]. issn 2175-9146. Available from doi:10.1590/jatm.v15.1297. 6. ZIPLINE. Zipline technology [online]. [N.d.]. [visited on 2023-06-09]. Available from: https://www. flyzipline.com/technology. 7. AEROSPACE, C-Astral. C-Astral Bramor ppX [online]. [N.d.]. [visited on 2023-06-09]. Available from: https://www.c-astral.com/es/sistemas-no-tripulados/bramor-ppx. 8. RAYMER, Daniel. Aircraft Design: A Conceptual Approach, Sixth Edition [online]. Washington, DC: American Institute of Aeronautics and Astronautics, Inc., 2018 [visited on 2023-06-06]. isbn 978-1-62410490-9. Available from doi:10.2514/4.104909. 9. SOROAGA, Petru. Ruby FPV website. 2023. Available also from: https://rubyfpv.com/. 10. Commercial drone market size, share trends analysis report by product, by application, by end-use, by propulsion type, by range, by operating mode, by endurance, by region, and segment forecasts, 2023 - 2030 [online]. 2021. [visited on 2023-06-13]. Tech. Rep. Report ID 978-1-68038-482-6. G. V. Research. Available from: https://www.grandviewresearch.com/industry-analysis/global-commercialdrones-market. 11. Shark Finesse - What is a good CAGR? [online]. [N.d.]. [visited on 2023-06-13]. Available from: https: //sharkfinesse.com/blog/what-is-a-good-cagr. 72 REFERENCES REFERENCES 12. UAV Drones Global Market Report 2023 [online]. 2023. [visited on 2023-06-13]. Tech. Rep. The Bussiness Research Companny. Available from: https://www.thebusinessresearchcompany.com/report/uavdrones-global-market-report. 13. KANE, Mark. Tesla’s 4680-Type Battery Cell Teardown: Specs Revealed [online]. 2022. [visited on 202306-06]. Available from: https://insideevs.com/news/598656/tesla-4680-battery-cell-specs/. 14. SU, Hai; CHEN, Zifeng; LI, Mengjie; BAI, Panxing; LI, Yong; JI, Xiao; LIU, Ziqiang; SUN, Jie; DING, Jia; YANG, Ming; YAO, Xiayin; MAO, Chong; XU, Yunhua. Achieving Practical High-Energy-Density Lithium-Metal Batteries by a Dual-Anion Regulated Electrolyte. Advanced Materials [online]. 2023, p. 2301171 [visited on 2023-06-06]. issn 0935-9648, issn 1521-4095. Available from doi: 10.1002/adma. 202301171. 15. AURAMBOUT, Jean-Philippe; GKOUMAS, Konstantinos; CIUFFO, Biagio. Last mile delivery by drones: an estimation of viable market potential and access to citizens across European cities. European Transport Research Review. 2019, vol. 11, no. 1, p. 30. issn 1866-8887. Available from doi: 10.1186/ s12544-019-0368-2. 16. TEAM, Zipline. Ghana Instant Delivery Supply Chain. Zipline Articles [online]. 2022 [visited on 2023-0703]. Available from: https://www.flyzipline.com/newsroom/stories/articles/ghana-instantdelivery-supply-chain. 17. KOROSEC, Kirsten. Zipline is now the national drone service provider for Rwanda. Tech Crunch [online]. 2022 [visited on 2023-07-03]. Available from: https://techcrunch.com/2022/12/15/zipline-isnow-the-national-drone-service-provider-for-rwanda/. 18. LYKOU, Georgia; MOUSTAKAS, Dimitrios; GRITZALIS, Dimitris. Defending Airports from UAS: A Survey on Cyber-Attacks and Counter-Drone Sensing Technologies. Sensors [online]. 2020, vol. 20, no. 12, pp. 35–37 [visited on 2023-06-13]. issn 1424-8220. Available from doi:10.3390/s20123537. 19. CORVO, Sypaq. Precision Payload Delivery System [online]. 2023. [visited on 2023-09-08]. Available from: https://corvouas.com.au/wp-content/uploads/CORVO-PPDS-image-change-compressedcompressed.pdf. 20. CORVO, Sypaq. Corvo Website [online]. 2023. [visited on 2023-09-08]. Available from: https:// corvouas.com.au/. 21. KIRILL, Ryabov. Light UAV Corvo PPDS in Ukraine [online]. August 29, 2023. [visited on 2023-09-08]. Available from: https://en.topwar.ru/224776-legkie-bpla-corvo-ppds-na-ukraine.html. 22. GOH, G.D.; AGARWALA, S.; GOH, G.L.; DIKSHIT, V.; SING, S.L.; YEONG, W.Y. Additive manufacturing in unmanned aerial vehicles (UAVs): Challenges and potential. Aerospace Science and Technology. 2017, vol. 63, pp. 140–151. issn 1270-9638. Available from doi: https://doi.org/10.1016/j.ast. 2016.12.019. 23. BHIMASENA RESEARCH, TECHNOLOGY, AND DEVELOPMENT, INDONESIA.; HADI, Ghozali Suhariyanto; VARIANTO, Rivaldy; INSTITUT TEKNOLOGI BANDUNG, INDONESIA.; TRILAKSONO, Bambang Riyanto; INSTITUT TEKNOLOGI BANDUNG, INDONESIA.; BUDIYONO, Agus; KONKUK UNIVERSITY, REPUBLIC OF KOREA. Autonomous UAV System Development for Pay73 REFERENCES REFERENCES load Dropping Mission. The Journal of Instrumentation, Automation and Systems [online]. 2015, vol. 1, no. 2, pp. 72–77 [visited on 2023-06-06]. issn 2288582X. Available from doi: 10.21535/jias.v1i2.158 . 24. (3DR), 3D Robotics. Pixhawk 1 board manufacturer specifications. 2013. Available also from: https: //docs.px4.io/main/en/flight_controller/pixhawk.html. 25. LENNON, Andy. R/C Model Aircraft Design. 1st. Air Age Media Inc., 1996. isbn 0-911295-40-2. 26. SIMONS, Martin. Model Aircraft Aerodynamics [online]. 3rd ed. Herts HP2 7ST, England: Argus Books, 1994 [visited on 2023-04-12]. isbn 1-85486-121-2. 27. Airfoil Tools - Website. [N.d.]. Available also from: tp://airfoiltools.com/. 28. XFLR5 Website. [N.d.]. Available also from: http://www.xflr5.tech/xflr5.htm. 29. SADRAEY, Mohammad H. Aircraft design: A systems engineering approach. Chichester: John Wiley and Sons, 2013. Aerospace Series. isbn 9781118352700. Available from doi:10.1002/9781118352700. 30. BIGAZZI, Luca; BASSO, Michele; BONI, Enrico; INNOCENTI, Giacomo; PIERACCINI, Massimiliano. A Multilevel Architecture for Autonomous UAVs. Drones [online]. 2021, vol. 5, no. 3, p. 55 [visited on 2023-09-05]. issn 2504-446X. Available from doi:10.3390/drones5030055. 31. TEAM, Ardupilot Development. Arduplane Complete Parameters List. 2023. Available also from: https://ardupilot.org/plane/docs/parameters.html. 74