How the initial level of trust in automated driving impacts drivers’ behaviour and early trust construction

Trust in Automation is known to influence human-automation interaction and user behaviour. In the Automated Driving (AD) context, studies showed the impact of drivers’ Trust in Automated Driving (TiAD), and linked it with, e.g., difference in environment monitoring or driver’s behaviour. This study investigated the influence of driver’s initial level of TiAD on driver’s behaviour and early trust construction during Highly Automated Driving (HAD). Forty drivers participated in a driving simulator study. Based on a trust questionnaire, participants were divided in two groups according to their initial level of TiAD: high (Trustful) vs. low (Distrustful). Declared level of trust, gaze behaviour and Non-Driving-Related Activities (NDRA) engagement were compared between the two groups over time. Results showed that Trustful drivers engaged more in NDRA and spent less time monitoring the road compared to Distrustful drivers. However, an increase in trust was observed in both groups. These results suggest that initial level of TiAD impact drivers’ behaviour and further trust evolution.     

Mapping drivers’ mental models of adaptive cruise control to performance

Driver support features (DSF) have the potential to improve safety, but they also change the driver-vehicle relationship —as well as their respective roles and responsibilities. To maximize safety, it is important to understand how drivers’ knowledge and understanding of these technologies—referred to as drivers’ mental models—impact performance and safety. This simulator study examined how drivers with different mental models of adaptive cruise control performed in edge cases. The study compared the responses of groups of drivers, with strong and weak mental models of ACC, established through a combination of screening, training, and exposure, in edge case situations in a high-fidelity driving simulator. In general, participants with strong mental models were faster than those with weak mental models to respond in edge-case situations—defined as cases where the ACC did not detect an approaching object, such as a slow-moving motorcycle. The performance deficits observed for drivers with weak mental models appear to reflect uncertainty surrounding how ACC will behave in edge cases.     

Driver conflict response during supervised automation: Do hands on wheel matter?

Securing appropriate driver responses to conflicts is essential in automation that is not perfect (because the driver is needed as a fall-back for system limitations and failures). However, this is recognized as a major challenge in the human factors literature. Moreover, in-depth knowledge is lacking regarding mechanisms affecting the driver response process. The first aim of this study was to investigate how driver conflict response while using highly reliable (but not perfect) supervised automation differ for drivers that (a) crash or avoid a conflict object and (b) report high trust or low trust in automation to avoid the conflict object. The second aim was to understand the influence on the driver conflict response of two specific factors: a hands-on-wheel requirement (with vs. without), and the conflict object type (garbage bag vs. stationary vehicle). Seventy-six participants drove with highly reliable but supervised automation for 30 min on a test track. Thereafter they needed to avoid a static object that was revealed by a lead-vehicle cut-out. The driver conflict response was assessed through the response process: timepoints for driver surprise reaction, hands-on-wheel, driver steering, and driver braking. Crashers generally responded later in all actions of the response process compared to non-crashers. In fact, some crashers collided with the conflict object without even putting their hands on the wheel. Driver conflict response was independent of the hands-on-wheel requirement. High-trust drivers generally responded later than the low-trust drivers or not at all, and only high trust drivers crashed. The larger stationary vehicle triggered an earlier surprise reaction compared to the garbage bag, while hands-on-wheel and steering response were similar for the two conflict object types. To conclude, crashing is associated with a delay in all actions of the response process. In addition, driver conflict response does not change with a hands-on-wheel requirement but changes with trust-level and conflict object type. Simply holding the hands on the wheel is not sufficient to prevent collisions or elicit earlier responses. High trust in automation is associated with late response and crashing, whereas low trust is associated with appropriate driver response. A larger conflict object trigger earlier surprise reactions.     

Emotional expressions facilitate human–human trust when using automation in high-risk situations

This study investigated the utility of emotional expression for human decision aids, when human aids conflict with an automated decision support system (DSS). The increasing presence of automation in society has resulted in critical, and often life threatening, situations when information from human and automated sources disagree. It has been known that reliance on human aids decrease during high-risk situations, while reliance on automated aids increase. However, it is also possible that human decision aids gain credibility from users when they embody the charismatic and emotionally expressive gesticulations seen in successful organizational leaders. The present study tested how a human agent's expressiveness when providing information would influence participants' behavioral reliance. Using the program Convoy Leader, participants (n=56) engaged in three decision-making scenarios where risk was manipulated as a within-subject factor and emotional expression as a between-subject factor. Emotional susceptibility, perceived risk, and trust for human as well as automated aids were measured. Overall trust was higher for the automated tool than human decision aid, and that pattern was amplified in conditions without an emotionally expressive human aid. Reliance was greater for emotionally expressive human aids, than stoic human aids, particularly during high risk conditions. The findings suggest that emotional expression of a human aid significantly impacts both reliance and trust of a decision aid, especially at higher risk levels. Emotionally expressive human agents should be utilized in decision conflicts where the automated system has certainly failed.     

Refreshing Older Adults’ Driving Skills (ROADSkills): A randomized controlled trial examining the effect of video feedback

ObjectivesThis study evaluated the effectiveness of a personalized video-based driver training program on the behind-the-wheel skills of community-dwelling older adults.MethodIn this randomized controlled trial (RCT), 80 older drivers [mean age (SD) = 71.0 (3.9)] were randomly assigned to view one of two educational videos: 1) personalized video feedback on their driving (n = 40) or 2) a generic video on aging-in-place (n = 40). The primary outcome, the total number of errors accrued in a 30-minute standardized on-road evaluation, was analyzed at baseline and 4 weeks after watching the assigned video. On-road evaluations were video-recorded and scored by a blinded rater. Self-report measures of driving abilities, behaviors, and comfort were also compared.ResultsAt the 4-week follow-up, the mean difference in the number of on-road performance errors [mean (95% CI) = -12.0 (-16.6, −7.4), p < 0.001] favored the intervention group compared to controls, with significant improvements across multiple domains: vehicle control [mean (95% CI) = -4.8 (-8.1, −1.5), p < 0.01], observing the roadway [mean (95% CI) = -5.5 (-8.4, −2.6), p < 0.001], and compliance with traffic rules [mean (95% CI) = -1.3 (-2.3, −0.2), p < 0.05]. Within-group change on behind-the-wheel errors was significant for the intervention group [mean (95% CI) = -10.3 (-13.5, −7.1), p < 0.001], but not for the control group [mean (95% CI) = 1.7 (-1.6, 4.9), p > 0.05]. No significant differences were found on the self-report measures.DiscussionA novel, video-based approach that provided personalized feedback to older drivers significantly reduced behind-the-wheel errors 4-weeks post-intervention. Changes in self-reported driving behaviors and abilities were not significant. Future research will examine the long-term effects of this approach on older adults’ actual and perceived driving abilities.     

The time windows of the sense of agency

The sense of agency depends on some internal cues that derive from action control, as well as external cues like contextual information and prior information (degree of contingency between an action and is effect). We assessed whether external agency cues are combined with internal agency cues to affect the sense of agency. In two experiments participants performed a movement (button press) that elicited, after a varying delay, an effect (ball appearing on a screen), and reported their sense of agency over the effect (full, partial or no-agency) while internal cues (premotor information) and external cues (contextual and prior information) were manipulated. We assessed the effect of agency cues on the delays at which the sense of agency varied. The delays were increased with premotor signals but were decreased with contextual information. These findings favour a model of integration of internal and external agency cues over time.