Multidimensional Assessment of Takeover Performance in Conditionally Automated Driving

Kexin Liang, Jan Luca Kästle, Bani Anvari, Simeon C. Calvert, J.W.C. van Lint · 2025

Read Multidimensional Assessment of Takeover Performance in Conditionally Automated Driving by Kexin Liang, Jan Luca Kästle, Bani Anvari, Simeon C. Calvert, J.W.C. van Lint (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-22252.txt.

Introduction

Kexin Liang and co-authors tested how well drivers take back control from a self-driving car in 2025. In mid-level automation, the car drives until a hard case forces a request for the driver to step in. That switch is a key safety point. The team scored takeovers in a driving simulator on three sides: speed of response, user experience, and safety of the drive.

They used a boosted tree model to weigh two traits of the driver. One is how well the driver grasps the scene, and the other is spare mental room. Scene grasp sped the first response most. Spare room shaped overall quality more, lifting both scores drivers gave and the paths they drove. The two traits play distinct, paired roles.

Catalog record

CREATORKexin Liang, Jan Luca Kästle, Bani Anvari, Simeon C. Calvert, J.W.C. van Lint
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE83 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-22252

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Multidimensional Assessment of Takeover Performance in Conditionally Automated Driving Kexin Liang, Jan Luca Kästle, Bani Anvari, Simeon C. Calvert, J.W.C. van Lint Abstract When automated driving systems encounter complex situations beyond their operational capabilities, they issue takeover requests, prompting drivers to resume vehicle control and return to the driving loop as a critical safety backup. However, this control transition places significant demands on drivers, requiring them to promptly respond to takeover requests while executing high-quality interventions. To ensure safe and comfortable control transitions, it is essential to develop a deep understanding of the key factors influencing various takeover performance aspects. This study evaluates drivers’ takeover performance across three dimensions: response efficiency, user experience, and driving safety— using a driving simulator experiment. EXtreme Gradient Boosting (XGBoost) models are used to investigate the contributions of two critical factors, i.e., Situational Awareness (SA) and Spare Capacity (SC), in predicting various takeover performance metrics by comparing the predictive results to the baseline models that rely solely on basic Driver Characteristics (DC). The results reveal that (i) higher SA enables drivers to respond to takeover requests more quickly, particularly for reflexive responses; and (ii) SC shows a greater overall impact on takeover quality than SA, where higher SC generally leads to enhanced subjective rating scores and objective execution trajectories. These findings highlight the distinct yet complementary roles of SA and SC in shaping performance components, offering valuable insights for optimizing human-vehicle interactions and enhancing automated driving system design. [orcid=0000-0002-8417-3974] keywords Situational awareness ,Spare capacity ,Reaction times ,Takeover qualities ,Conditionally automated driving † † corresponding: Corresponding author at: Transport and Planning Department, Faculty of Civil Engineering and Geosciences, Delft University of Technology. 1 Introduction Automated driving systems continue to advance but still face limitations in handling complex and risky driving scenarios. When these systems reach their operational boundaries, in many cases they issue a takeover request, prompting drivers to resume manual control of the vehicle within constrained time budgets. Ensuring smooth vehicle control transitions is crucial for avoiding potential hazards and providing a comfortable driving experience. Control transitions are challenging because they require drivers to timely shift from a state of passive monitoring or non-driving activities to active vehicle control ( Hu et al. 2023 ; Lu et al. 2016 ) . This shift is not instantaneous; it demands time and cognitive resources to assess the situation, make decisions, and act. If drivers do not take over vehicle control promptly enough or if their takeover responses are poor in quality, the risks of accidents increase, and the overall user experience deteriorates. Thus, understanding the factors that influence drivers’ takeover performance during control transitions and accordingly adopting necessary interventions are essential for enhancing the safety and comfort of conditionally automated driving. A widely acknowledged factor influencing drivers’ takeover performance is Situational Awareness (SA), which reflects how well a driver perceives, understands, and projects the surrounding environment ( Endsley 2020 ; Van Lint and Calvert 2018 ) . Studies by van den Beukel and van der Voort 2013 and Tan and Zhang 2022 demonstrated that drivers with higher SA levels had greater success in executing safe takeover maneuvers. Vlakveld et al. 2018 found that drivers’ SA was closely related to their hazard perception, suggesting that drivers with better SA are more capable of recognizing potential risks, and consequently, reducing crash rates. McKerral et al. 2023 pointed out that improved SA enhanced takeover performance by guiding drivers’ responses to takeover requests. Although the positive relationship between SA and takeover performance is well-documented, the concept of “takeover performance” is multifaceted and encompasses various aspects of driver behaviours and responses during control transitions ( Li et al. 2023 ) . Notably, certain aspects of takeover performance can even conflict: for instance, a shorter takeover time does not always indicate better quality ( Wu et al. 2022 ; Wang et al. 2022 ) . While a shorter time to resume meaningful human control of the vehicle ( Calvert et al. 2024 ) refers to a more efficient takeover, it may come at the cost of comfort if achieved through abrupt actions, such as hard braking, and can also pose pressure on following vehicles to respond effectively which increases crash risks ( Wu et al. 2022 ) . Furthermore, discrepancies between drivers’ subjective evaluations of their takeover quality and objective metrics based on driving trajectory have been observed ( Guo et al. 2021 ) , though not well studied. These complexities underscore the need for a more detailed understanding of how SA affects various dimensions of takeover performance. In addition to SA, this study investigates another critical factor influencing takeover performance: Spare Capacity (SC), derived from Task-Capability Interface (TCI) theory ( Fuller 2005 ) . TCI theory suggests that drivers adjust their behaviours based on the dynamic interactions between their perceived task capability (TC) and task demands (TD) to maintain their perceived safety margin at an acceptable level ( Fuller 2005 ) . On this basis, SC is defined as TC minus TD ( Fuller 2000 ) , inversely capturing how difficult a scenario is for a driver to manage. As Fuller 2011 noted, drivers with lower SC were more vulnerable to performance errors and the challenges of high-demand situations. In the context of automated driving, Müller et al. 2021 used drivers’ secondary task performance as an indicator of SC, finding that drivers engaged in high-workload secondary tasks exhibited reduced SC and required longer reaction times. Similarly, our previous work ( Liang et al. 2024 ) found that drivers with higher SC levels during takeover tasks generally resumed conscious manual control of the vehicle more quickly. We argue that SC has the potential to influence other aspects of takeover performance, especially takeover qualities, but its impacts require further investigation. This study aims to systematically investigate the effects of SA and SC on various aspects of takeover performance. A driving simulator experiment is conducted, where drivers are instructed to take over vehicle control from automation across nine scenarios (three traffic densities * three non-driving related tasks). For each takeover, drivers’ SA and SC are measured using both (i) questionnaires that are adopted from well-established instruments for takeover contexts, and (ii) cognition-related eye movement metrics. Meanwhile, takeover performance is systematically evaluated along three dimensions: (i) drivers’ reaction times, (ii) subjective takeover qualities, and (iii) objective takeover qualities. With this framework, we conduct a comprehensive analysis of the effects of drivers’ SA and SC on various aspects of takeover performance, which is the main contribution of this study. This study can provide valuable insights to readers who are interested in drivers’ cognition activities during control transitions and their implications for enhancing human-vehicle interactions for safe and comfortable conditionally automated driving. The remainder of this paper is structured as follows: Section 2 details the conducted driving simulator experiment, the questionnaire for measuring drivers’ cognition and subjective evaluations, the collected data, and the takeover performance prediction models; Section 3 presents the prediction results and the quantified effects of situational awareness and spare capacity on various aspects of takeover performance; Section 4 discusses the results and deliberates on the limitations of this study; and Section 5 summarizes the findings of this study and gives recommendations for future research. 2

Read the complete text