Fabienne Roche
Assessing subjective criticality of take-over
situations: Validation of two rating scales
Open Access via institutional repository of Technische Universität Berlin
Document type
Journal article | Accepted version
(i. e. final author-created version that incorporates referee comments and is the version accepted for
publication; also known as: Author’s Accepted Manuscript (AAM), Final Draft, Postprint)
This version is available at
https://doi.org/10.14279/depositonce-16389
Citation details
Roche, F. (2021). Assessing subjective criticality of take-over situations: Validation of two rating scales. In
Accident Analysis amp; Prevention (Vol. 159, p. 106216). Elsevier BV.
https://doi.org/10.1016/j.aap.2021.106216.
Terms of use
This work is protected by copyright and/or related rights. You are free to use this work in any way permitted by
the copyright and related rights legislation that applies to your usage. For other uses, you must obtain
permission from the rights-holder(s).
ASSESSING SUBJECTIVE CRITICALITY
1
Title: Assessing subjective criticality of take-over situations: Validation of two rating scales 1
Author: Fabienne Roche
2
Corresponding author: Fabienne Roche, [email protected] 3
Address: Technische Universität Berlin 4
Fachgebiet Kognitionspsychologie und Kognitive Ergonomie 5
MAR 3-2 6
Marchstraße 23 7
10587 Berlin, Germany 8
Running head: Assessing Subjective Criticality 9
Manuscript type: Research paper 10
Declarations of interests: none 11
Funding sources involved: This work has resulted from the interdisciplinary research project 12
“Analysis and Support of Driver Interventions in Dynamic: Critical Situations during Highly 13
Automated Driving” funded by the German Research Community (Deutsche 14
Forschungsgemeinschaft, DFG, Grant No. 326727090). 15
ASSESSING SUBJECTIVE CRITICALITY
2
Abstract 16
Assessing subjective criticality of take-over situations is crucial for understanding of take-over 17
behavior and comparing studies. However, no validated rating scales exist that assess subjective 18
criticality of take-over situations. In a driving simulator study, two rating scales, the Scale of 19
Criticality Assessment of driving situations from Neukum et al. (2008) and the Criticality Rating 20
Scale, were tested on their validity to assess the subjective criticality of take-over situations. 21
Besides, the subjective and behavioral changes over the repeated experience of take-over 22
situations were investigated. Twenty-five participants experienced a set of five take-over 23
situations with varying time-to-collisions (TTC) at the moment of the take-over request, twice. 24
After each of the first five take-over situations, participants rated the criticality on one scale, after 25
each of the second five situations on the other scale. Correlation coefficients between TTCs and 26
criticality ratings for each scale were calculated. Also, the changes of subjective and behavioral 27
measures over the trials were investigated. Correlation coefficients indicated a strong correlation 28
between criticality ratings and TTCs. Hence, both scales are equally valid for the assessment of 29
the criticality of take-over situations. The repeated experience of the take-over situations did not 30
affect effort ratings, take-over times, or steering wheel positions. But brake input decreased with 31
increasing practice, indicating a safer take-over behavior. Hence, results of studies with repeated 32
experience of take-over situations are relatively valid as only brake behavior changed with 33
increasing practice. 34
Keywords: Automated driving, Driver-vehicle interaction, Driver behavior, Criticality, Take-35
over behavior, Scale validation 36
ASSESSING SUBJECTIVE CRITICALITY
3
1.1 Introduction 37
In the past years, human factors researchers have accumulated an impressive amount of 38
knowledge, especially on the take-over process (Gold et al., 2016; Jamson et al., 2013; Körber et 39
al., 2016; Murata et al., 2013; Politis et al., 2014; Roche et al., 2018; SAE International, 2018). It 40
was observed that different characteristics of take-over situations may heavily influence the take-41
over behavior and subjective experience (Damböck et al., 2012; Gold et al., 2016, 2013; 42
Radlmayr et al., 2014; Roche & Brandenburg, 2020, 2018). One of these characteristics is the 43
objective criticality of the take-over situation which is determined by situational parameters. For 44
example, take-over situations with low time budgets are more critical than situations with high 45
time budgets. The objective criticality affects the take-over behavior and subjective criticality, 46
e.g. more extreme behavior and higher subjective criticality when the situation is more critical. 47
There are many options to assess take-over behavior, such as take-over times or steering 48
behavior. It provides insights into how drivers behave depending on different situational 49
parameters. In contrast, to our knowledge, no validated instrument exists to assess the subjective 50
criticality, even though, it supports the interpretation of observed take-over behavior and may 51
enable comparisons between different take-over situations. Therefore, in the present study, two 52
rating scales are validated regarding their suitability to assess the subjective criticality of take-53
over situations. Besides, subjective and behavioral changes of the repeated experience of take-54
over situations are investigated. 55
1.2 Objective Criticality of Driving Situations 56
The objective criticality of a driving situation is ‘the accident risk’ (Rodemerk et al., 57
2012, p. 1). Hence, a driving situation, in which a collision is inevitable, constitutes the highest 58
possible objective criticality (Rodemerk et al., 2012). Especially in automated driving, the 59
ASSESSING SUBJECTIVE CRITICALITY
4
objective criticality of a driving situation is crucial since it influences the take-over behavior 60
(Gold et al., 2013; Roche & Brandenburg, 2020, 2018; Zhang et al., 2019). 61
The objective criticality of a driving situation may be determined by situational 62
parameters such as time budget (Junietz et al., 2017), traffic density, or visibility. Lower time 63
budgets, higher traffic densities, or poor visibility may lead to a higher objective criticality. In 64
this paper, we focus on time budget. It can be quantified by time-to-collision (TTC). TTC 65
describes the available time until a vehicle would collide with a reference object (Vogel, 2003). A 66
reference object may be a preceding vehicle or a system boundary, such as an obstacle on the 67
road. Hence, shorter TTCs indicate a more critical situation. These more critical situations may 68
emerge in case the automated driving system reaches its limits or in the case of driver-initiated 69
take-overs (Roche et al., 2020). 70
TTC is known to influence take-over behavior. Numerous driving simulator studies varied 71
the TTC in take-over situations. Lower TTCs, hence more critical take-over situations, were 72
associated with lower take-over times (Gold et al., 2013; Roche & Brandenburg, 2020, 2018; 73
Zhang et al., 2019), higher decelerations (Roche et al., 2020; Roche & Brandenburg, 2020, 2018), 74
and larger steering wheel angles (Roche et al., 2020; Roche & Brandenburg, 2020, 2018). While 75
lower take-over times are a desirable behavior, high decelerations and extreme steering are a 76
threat to the drivers’ safety for the following reasons: This behavior may result in (a) vehicle 77
instability, (b) rear-end collisions with following vehicles, (c) collisions with vehicles on 78
neighboring lanes or (d) lane departures. Indeed, more critical take-over situations in terms of 79
lower TTCs led to higher error rates, such as collisions or missing lane changes (Damböck et al., 80
2012), more lane departures (Mok, Johns, Lee, Ive, et al., 2015; Mok, Johns, Lee, Miller, et al., 81
Loading more pages...