Source-linked AI summary
Examining the Effects of Emotional Valence and Arousal on Takeover Performance in Conditionally Automated Driving
Na Du, Feng Zhou, Elizabeth Pulver, Dawn M. Tilbury, Lionel P. Robert, Anuj K. Pradhan, X. Jessie Yang
TL;DR
Drivers can struggle to resume control during conditionally automated driving, and the role of emotion in these transitions has been understudied. Using a 32-participant driving-simulation experiment with movie-clip emotion induction, the study measured takeover time and quality. Positive valence improved takeover quality, whereas high arousal did not improve takeover time.
Problem
Little is known about how emotional valence and arousal influence takeover performance in conditionally automated driving, despite emotion’s role in human-machine interaction and manual driving.
Method
A driving-simulation experiment with 32 participants used movie clips to induce emotion and measured takeover time and quality during automated driving.
Results
Positive valence improved takeover quality through smaller maximum resulting acceleration and jerk, while high arousal did not improve takeover time.
Takeaways & Limitations
Benefits of positive emotions carried over from manual driving to conditionally automated driving, whereas benefits of arousal did not.
Takeaways & Limitations
The study used a fixed-based desktop driving simulator with limited fidelity.
Abstract
from arXiv · showhide
In conditionally automated driving, drivers have difficulty in takeover transitions as they become increasingly decoupled from the operational level of driving. Factors influencing takeover performance, such as takeover lead time and the engagement of non-driving related tasks, have been studied in the past. However, despite the important role emotions play in human-machine interaction and in manual driving, little is known about how emotions influence drivers takeover performance. This study, therefore, examined the effects of emotional valence and arousal on drivers takeover timeliness and quality in conditionally automated driving. We conducted a driving simulation experiment with 32 participants. Movie clips were played for emotion induction. Participants with different levels of emotional valence and arousal were required to take over control from automated driving, and their takeover time and quality were analyzed. Results indicate that positive valence led to better takeover quality in the form of a smaller maximum resulting acceleration and a smaller maximum resulting jerk. However, high arousal did not yield an advantage in takeover time. This study contributes to the literature by demonstrating how emotional valence and arousal affect takeover performance. The benefits of positive emotions carry over from manual driving to conditionally automated driving while the benefits of arousal do not.
INTRODUCTION
Conditionally automated vehicles require drivers to resume control at short notice, yet drivers struggle during takeover transitions. This study examines how emotional valence and arousal affect takeover performance.
- INTRODUCTION: Drivers in conditional automation may engage in non-driving tasks but must quickly resume vehicle control when automation reaches its limits.The transition includes terminating non-driving tasks, returning eyes, hands, and feet to driving-related positions, and resuming control.
- INTRODUCTION: Takeover transitions are difficult as drivers become increasingly decoupled from the operational level of driving.
- INTRODUCTION: Prior research has examined environmental conditions, non-driving tasks, individual characteristics, and human-machine-interface design as influences on takeover performance.
- INTRODUCTION: Despite emotion’s role in human-machine interaction and manual driving, little is known about its influence on takeover performance.
- INTRODUCTION: The study examines emotional valence and arousal systematically in conditionally automated driving.The dimensional approach treats valence and arousal as distinct emotion dimensions rather than focusing on discrete emotions.
Emotions in Manual and Automated Driving
Emotion affects manual driving and human-machine interaction, but evidence for conditionally automated driving remains limited. Prior findings associate positive valence with vehicle control and higher arousal with faster hazard responses, although results vary across induction methods and tasks.
- Emotions in Manual and Automated Driving: Elevated emotional states were associated with a 9.8-times higher crash risk in naturalistic driving data.
- Emotions in Manual and Automated Driving: Manual-driving studies generally linked positive valence with better vehicle control, but some music-induced results showed worse lateral control under positive valence.
- Emotions in Manual and Automated Driving: Higher arousal was associated with faster hazard responses and improved responsiveness to followed-vehicle speed changes in manual-driving tasks.
- Emotions in Manual and Automated Driving: Few studies have examined how emotions influence driving performance in conditionally automated driving, and existing work has primarily focused on detecting emotional states.
- Emotions in Manual and Automated Driving: Emotion may influence takeover transitions through drivers’ perception, cognitive processing, and decision making, but the paper reports no prior studies directly examining this topic.
The Present Study
The study tests how valence and arousal affect takeover quality and response time in conditional automation. It predicts benefits of positive emotions for takeover quality but reduced or absent arousal benefits for response time.
- The Present Study: The study examines how emotions affect drivers’ takeover performance in conditionally automated driving.
- The Present Study: Positive emotions were hypothesized to enhance takeover quality through broader attention and thought-action repertoires supporting situation comprehension and action selection.
- The Present Study: High arousal was hypothesized to provide reduced or no advantage in takeover response time because drivers can respond reflexively to takeover requests.
- The Present Study: A human-subject experiment used 32 drivers, a fixed-based simulator, Level 3 automation, movie clips for emotion induction, and four takeover events.Takeover time and quality were recorded and analyzed.
METHOD
The study complied with the American Psychological Association code of ethics and received Institutional Review Board approval from the University of Michigan.
- METHOD: The research followed the American Psychological Association code of ethics and received University of Michigan Institutional Review Board approval.
Participants
The study included 32 university students with valid U.S. driver licenses and normal or corrected-to-normal vision, using a fixed-based desktop simulator programmed to represent SAE Level 3 automation.
- Participants: 32 university students participated, averaging 21.4 years of age; 17 were female and 15 were male.Participants were screened for valid U.S. driver licenses and susceptibility to simulator sickness.
- Participants: The experiment lasted 60–75 minutes, and participants received $30 upon completion.
- Apparatus: A fixed-based desktop simulator displayed forward road scenes on a 32-inch monitor and rear-view imagery in a separate window.Vehicle control used a Logitech steering wheel and pedal system connected to the simulator.
- Apparatus: The simulator represented SAE Level 3 automation, with the automated vehicle performing longitudinal and lateral control until unexpected events required driver takeover.
Experimental Design
The within-subjects experiment manipulated emotional valence and arousal through movie clips before standardized takeover events, with condition order counterbalanced across participants.
- Emotion induction: Eight four-minute movie clips induced emotional conditions spanning the four quadrants of the valence–arousal space.The clips were selected based on prior literature and used to manipulate emotional valence and arousal within participants.
- Experimental control: A Latin square design counterbalanced the sequence of emotion-induction conditions to reduce ordering effects.
- Takeover events: Participants watched the movie clips while the automated vehicle was active and were not required to monitor the driving environment until the takeover request.The request combined an audible “takeover” phrase with a red-hands steering-wheel icon, presented consistently across participants.
- Takeover events: Each participant experienced four takeover events in which the automated vehicle issued a takeover request four seconds before a construction zone, bicyclist, police vehicle, or swerving vehicle.Participants were expected to change from the right lane to the left lane during the transition.
Measures
Takeover performance was evaluated through subjective emotion ratings, takeover timing, and objective measures of maneuver quality and collision risk.
- Emotion measures: The Self-Assessment Manikin measured emotional valence from 1, extremely negative, to 9, extremely positive, and arousal from 1, lowest, to 9, highest.
- Timing measures: Takeover time was measured from the takeover request to the start of the maneuver, defined by a 2-degree steering change or 10% pedal depression.
- Quality measures: Takeover quality was assessed using maximum resulting acceleration, maximum resulting jerk, and minimum time to collision during lane changing.The assessment window extended from the takeover request until the vehicle reached the boundary of the other lane.
- Quality measures: Smaller resulting acceleration and jerk represented smoother, safer, and higher-quality takeover reactions.Jerk is the derivative of acceleration and was also used to evaluate ride comfort and driving aggressiveness.
- Data handling: Five crashes occurred, so minimum time to collision was treated as not applicable in those cases.Other driving-dynamic variables were calculated using the interval from the takeover request to driver re-engagement.
Experimental procedure
After training on manual control and automation engagement, participants watched emotion-inducing clips during automated driving and immediately responded to takeover requests before completing emotion ratings.
- Training: Participants first completed a 5-minute training session covering lane changes and engaging or disengaging automated driving with a steering-wheel button.They practiced an unexpected takeover while watching a one-minute Zen Garden clip.
- Driving instructions: Participants were instructed to obey traffic laws and drive at the 35-mph speed limit during manual control.
- Experimental procedure: During each course, participants activated automated driving, watched two four-minute movie clips, and received a takeover request near the end of the clips.
- Takeover procedure: Participants immediately took control, negotiated the driving scenario, and then handed control back to the automated vehicle.The scenarios included traffic lights that did not work and required drivers to observe their surroundings.
- Post-takeover assessment: After re-engaging automation, participants recalled the clips and completed the Self-Assessment Manikin survey for emotional valence and arousal.
RESULTS
Positive emotional valence improved takeover quality, producing smaller maximum resulting acceleration and jerk, while valence and arousal did not significantly affect takeover timeliness.
- Takeover quality: Positive valence led to smaller maximum resulting acceleration, indicating better takeover quality.The valence effect was significant (F (1, 56) = 4.26, p = .04), and Figure 6 presents the comparison.
DISCUSSION
Positive emotional valence improved takeover quality, whereas high arousal did not improve takeover time in conditionally automated driving. The findings suggest that some benefits observed in manual driving carry over to takeover transitions, but the broad metric variability complicates comparisons across studies.
- DISCUSSION: Positive valence improved takeover quality, reflected in smaller maximum resulting acceleration and maximum resulting jerk.These metrics are associated with safety, ride comfort, and lower driving aggressiveness.
- DISCUSSION: The benefits of positive valence in manual driving can carry over to conditionally automated driving.The authors relate this result to prior manual-driving findings in which positive valence improved vehicle control.
- DISCUSSION: Interpretation across takeover-quality studies remains difficult because prior research uses a wide range of crash, velocity, acceleration, jerk, and TTC metrics.The authors identify a need to examine whether standardized metrics for takeover quality can be proposed.
- DISCUSSION: High arousal did not improve takeover time, consistent with a non-significant effect on response time during takeover transitions.Drivers switched attention from the hands-free movie task to the driving scene in less than 2 seconds regardless of emotional state.
- DISCUSSION: The absence of an arousal advantage may reflect that takeover requests direct attention to driving and the hands-free task can be abandoned immediately.Unlike some manual-driving hazard-detection tasks, the transition did not require physically ending the non-driving task or putting down a device.
- DISCUSSION: High arousal showed a trend toward smaller maximum resulting acceleration, suggesting a possible takeover-quality benefit that requires further research.The authors link this trend to arousal-related mobilization of attentional resources and effort for immediate actions.
CONCLUSION
This study shows that emotional valence and arousal influence takeover performance in conditionally automated driving, but their effects do not simply mirror manual-driving findings.
- CONCLUSION: The findings provide empirical evidence that emotion affects both takeover time and quality in conditionally automated driving.The study systematically manipulated emotional states to examine valence and arousal during takeover transitions.
- CONCLUSION: Positive emotions improved takeover quality, whereas arousal did not improve takeover time.The study reports smaller maximum resulting acceleration and jerk under positive valence, while high arousal provided no takeover-time advantage.
- CONCLUSION: Emotion findings from manual driving cannot simply be applied to automated-driving takeovers.Positive-emotion benefits carried over, but arousal benefits did not.
- CONCLUSION: Emotion-regulation systems could detect negative emotional states and use reappraisal or distraction to support takeover performance.The paper also suggests that an automated vehicle could delay or avoid handing over control when appropriate.
- CONCLUSION: The study was limited to one hazard-related lane-change takeover event in a fixed-base desktop simulator.Future research should examine other takeover events and higher-fidelity or naturalistic driving environments.
- CONCLUSION: Post-takeover emotion reports may have been influenced by the takeover event itself.Physiological measures collected before takeover and eye tracking were proposed as future alternatives for assessing emotion and attention.