Sentry: An Innovative Limb-Tracking device

Sentry: An Innovative Limb-Tracking Device Claudia Presicci Department of Informatics, Bioengineering, Robotics, and Systems Engineering, University of Genoa, SWHARD S.r.l. Genoa, Italy https://orcid.org/00090000-3981-3285 Paolo Dondero SWHARD S.r.l. Genoa, Italy https://orcid.org/00000002-9185-9413 Igor Ingegnosi SWHARD S.r.l. Genoa, Italy https://orcid.org/00090008-6372-0474 Ronny Stanzani SWHARD S.r.l. Genoa, Italy https://orcid.org/00000001-8649-7270 Abstract— Sentry is a modular tracking platform based on IMUs that unifies acquisition, real-time supervision, and offline kinematic reconstruction for human movement analysis. A Nordic-based controller orchestrates up to four BNO085 sensors with on-device preprocessing (quaternions, angular velocities, and linear accelerations), streaming via BLE or logging to microSD. A companion app enables immediate quality control; C/C++/Python tools reconstruct joint angles. Keywords—IMU, kinematics, wearable, human movement I. INTRODUCTION Wearable devices based on inertial measurement units (IMUs) enable the study of human movement beyond traditional laboratories, opening opportunities in clinical practice [1], [2], sports performance [3]–[5], and daily-life monitoring [6]–[8]. Despite this potential, real-world deployments still struggle with multi-sensor coordination, robust wireless streaming, and accessible software that spans both real-time supervision and offline analysis. To address these challenges, we developed Sentry [9], a modular IMUbased tracking platform that integrates hardware, firmware, and software into a cohesive pipeline for capturing, monitoring, and reconstructing kinematics in practical settings. Sentry has already undergone validation [10] and has been deployed in prior analyses, demonstrating its suitability for research-grade, out-of-lab movement assessment [11], [12]. In the present study, we specifically examine knee kinematics during gait, using Sentry to acquire and reconstruct joint-level trajectories (Fig. 1). Our aims are to provide a body-worn sensing system with flexible sensor configurations, ensure reliable low-latency telemetry for immediate quality control, and deliver a transparent offline toolchain for reconstructing joint angles from IMU orientations. Fig. 1. Subject wearing Sentry with four connected sensors (blue modules on elastic bands at thighs and shanks). II. METHODS A. System architecture Sentry comprises a Main Control Unit built around a Nordic Semiconductor System-on-Chip (SoC) [13] and up to four Bosch BNO085 IMUs [14] connected via I²C. Custom firmware outputs a configurable set of preprocessed signals, including quaternions, linear accelerations, angular velocities, and magnetic-field perturbations, up to 50 Hz. Data can be stored locally on microSD or streamed via Bluetooth Low Energy (BLE) to companion software. Battery life supports up to 12 hours of continuous acquisition. B. Software A Flutter-based Android application allows connection to one or more Sentry units over Bluetooth Low Energy (BLE), live reception of multi-sensor streams, and on-screen visualization of quaternion data in numerical and graphical forms. This also facilitates immediate verification that all sensors are correctly connected and positioned on the limb of interest. Offline analysis is handled by a combined C/C++ and Python toolchain. The pipeline imports the logged or streamed data, performs cleaning and basic calibration, and reconstructs joint angles from the orientation time series. It supports strideor repetition-level segmentation and produces summary statistics suitable for downstream comparisons. The processing steps are intentionally exposed, filtering parameters, calibration routines, drift handling, to facilitate troubleshooting and to maintain scientific transparency. C. Proof-of-concept evaluation To illustrate feasibility, we applied Sentry to a gait analysis example focused on knee flexion . Orientation data were collected during level walking and converted to jointangle trajectories across successive gait cycles. We examined the temporal organization and within-session variability to gauge whether the end-to-end system behaved coherently. Fig. 2. Screenshot of the mobile application connected to the device with four IMUs. Quaternions are displayed in both graphical and numerical form. 2025 I-RIM Conference October 17-19, Rome, Italy ISBN: 9788894580570 10.5281/zenodo.17629876 237 III. RESULTS AND DISCUSSION The system achieved stable multi-sensor streaming at up to 50 Hz with responsive visualization on the mobile interface (Fig. 2), allowing operators to detect misplacements or loose attachments in real time. In our tests, we ran uninterrupted ~4hour acquisition sessions via microSD logging. Fig. 3. Reconstruction of the right (red) and the left (blue) knee flexion of a subject during walking. In the gait example, the reconstructed knee flexion demonstrated the expected structure over stance and swing, including a clear swing-phase peak and consistent timing across cycles. Within-session dispersion of the flexion trajectory was low, suggesting that synchronization, orientation estimation, and attachment stability were adequate for joint-level reconstruction in this scenario. The results indicate that a pragmatic combination of on-device preprocessing and dual data paths can reduce common points of failure in field deployments. Delivering orientations directly from the sensors lowers bandwidth demands and simplifies the analysis pipeline for non-expert users. The realtime layer serves as an effective guardrail: by revealing quality issues during acquisition, it reduces wasted trials and shortens setup time. The offline toolchain closes the loop by converting orientations into interpretable kinematic variables through transparent steps, supporting reproducibility and domainspecific customization. The flexibility of the architecture is central to its practical value. Operators can tailor the number and placement of sensors to the protocol, from a minimal single-joint configuration to more complex multi-segment layouts. Because the system unifies acquisition, monitoring, and analysis, it can be inserted into clinical or sports workflows without requiring specialized motion-capture infrastructure. The present evaluation is preliminary in scope. The gait demonstration involves a single subject and task and does not constitute a formal validation against laboratorygrade systems. IV. CONCLUSIONS Sentry provides an end-to-end approach to IMU-based movement analysis that is tailored for real-world use. By combining a Nordic-based controller with up to four BNO085 sensors, configurable on-device preprocessing, dual streaming and logging pathways, a user-friendly mobile interface for real-time supervision, and an open offline reconstruction stack, the platform translates inertial sensing into a usable and reproducible workflow. This findings motivate a comprehensive validation of gait analysis including all lowerlimb and pelvic joints, with structured comparisons against optical motion capture across speeds and conditions, reporting absolute agreement and measurement reliability across joints and phases. Ongoing development targets two further directions. First, we plan to release activity-specific modules within the mobile app – for posture monitoring, rehabilitation, sportive performance analysis – supporting free sensor placement followed by calibration routines to resolve segment alignment; these modules will guide acquisition in real time and generate harmonized outputs (joint-angle waveforms, peak angles/ROM, spatiotemporal parameters) to enable rigorous cross-study comparability. Second, we aim to improve algorithmic features, e.g. drift-aware strategies without magnetometer, and adaptive sensor-fusion parameters that respond to context while keeping the user workflow simple. V. REFERENCES [1] F. Bo et al., “IMU-Based Monitoring for Assistive Diagnosis and Management of IoHT: A Review,” Healthcare (Basel), vol. 10, no. 7, p. 1210, June 2022, doi: 10.3390/healthcare10071210. [2] C. Gu, W. Lin, X. He, L. Zhang, and M. Zhang, “IMU-based motion capture system for rehabilitation applications: A systematic review,” Biomimetic Intelligence and Robotics, vol. 3, no. 2, p. 100097, June 2023, doi: 10.1016/j.birob.2023.100097. [3] C. Presicci, I. Ingegnosi, P. Dondero, M. Casadio, and C. Pierella, “IMU-Based Systems For Skiing Sports: A Scoping Review,” IEEE Access, vol. 13, pp. 152430–152451, 2025, doi: 10.1109/ACCESS.2025.3604571. [4] T. Kuwahara, H. Kawamoto, and Y. Sankai, “IMU Sensor Module for the Measurement of High-speed Motion in the Analysis of Human Skills,” in 2019 IEEE/SICE International Symposium on System Integration (SII), Jan. 2019, pp. 560–565. doi: 10.1109/SII.2019.8700421. [5] V. Mudeng, I. M. Hakim, S. S. Suprapto, and S. Choe, “An Alternative Athlete Monitoring System Using Cost-Effective Inertial Sensing Instrumentation,” J. Electr. Eng. Technol., vol. 17, no. 6, pp. 3581– 3592, Nov. 2022, doi: 10.1007/s42835-022-01258-1. [6] A. Ayman, O. Attalah, and H. Shaban, “Smart System for Recognizing Daily Human Activities Based on Wrist IMU Sensors,” in 2019 International Conference on Advances in the Emerging Computing Technologies (AECT), Feb. 2020, pp. 1–6. doi: 10.1109/AECT47998.2020.9194154. [7] F. M. Rast and R. Labruyère, “Systematic review on the application of wearable inertial sensors to quantify everyday life motor activity in people with mobility impairments,” Journal of NeuroEngineering and Rehabilitation, vol. 17, no. 1, p. 148, Nov. 2020, doi: 10.1186/s12984020-00779-y. [8] M. Martínez-Zarzuela, J. González-Alonso, M. Antón-Rodríguez, F. J. Díaz-Pernas, H. Müller, and C. Simón-Martínez, “Multimodal video and IMU kinematic dataset on daily life activities using affordable devices,” Sci Data, vol. 10, no. 1, p. 648, Sept. 2023, doi: 10.1038/s41597-023-02554-9. [9] “Sentry | Swhard - Cuttin’ Edge Technology.” Accessed: Sept. 23, 2025. [Online]. Available: https://www.swhard.it/sentry.php [10] R. Stanzani, P. Dondero, A. Mantero, and M. Testa, “Measurement Accuracy of an Upper Limb Tracking System Based on Two Hillcrest Labs BNO080 IMU Sensors: An Environmental Assessment,” IEEE Sensors Journal, vol. PP, pp. 1–1, May 2020, doi: 10.1109/JSEN.2020.2992733. [11] M. Job, S. Battista, R. Stanzani, A. Signori, and M. Testa, “Quantitative Comparison of Human and Software Reliability in the Categorization of Sit-to-Stand Motion Pattern,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. PP, pp. 1–1, Apr. 2021, doi: 10.1109/TNSRE.2021.3073456. [12] M. Iurato, R. Stanzani, M. Job, A. Gaggero, I. Ingegnosi, and M. Testa, “Assessment of Upper Limb Functional Workspace through Inertial Measurement Units: a Pilot Study,” in Human Interaction and Emerging Technologies (IHIET-FS 2025): Future Systems and Artificial Intelligence Applications, AHFE Open Acces, 2025. doi: 10.54941/ahfe1005965. [13] N. Semiconductor, “Nordic Semiconductor | Specialists in Low Power Wireless.” Accessed: Sept. 23, 2025. [Online]. Available: https://www.nordicsemi.com/ [14] “Ceva BNO 9-Axis IMU Precision Motion Tracking,” Ceva. Accessed: Sept. 23, 2025. [Online]. Available: https://www.cevaip.com/product/bno-9-axis-imu/ 238