Driver trust & mode confusion in an on-road study of level-2 automated vehicle technology. (October 2020)

• Record Type: Journal Article
• Authors: Wilson, Kyle M.; Yang, Shiyan; Roady, Trey; Kuo, Jonny; Lenné, Michael G
• Is Part Of: Safety science. Volume 130(2020)
• Publication Date: 2020-10
• Subjects: Automated driving -- Human-automation interaction -- Driver behavior -- Driver safety -- Human factors; Industrial accidents -- Periodicals; Accident Prevention -- Periodicals; Safety -- Periodicals; Travail -- Accidents -- Périodiques; 363.11
• DOI: 10.1016/j.ssci.2020.104845
• Languages: English
• ISSNs: 0925-7535

Abstract

Highlights: Acceptance for automated technology increases following on-road experience. Engagement is lower for partially automated driving than manual driving. The design of automated vehicle interfaces can cause driver confusion between modes. Drivers use strategies to counter hands-on steering wheel sensors. Complacent behaviours and attitudes threaten the safe use of automated vehicles. Abstract: Contextual investigations of automated vehicle technology have so far been rare, however they are crucial to uncover the challenges that exist around its acceptance and safe use. Twenty-one drivers used a partially automated vehicle on a public highway in unaltered traffic conditions, while their behaviour was observed. Subjective measures of technology acceptance were taken before and after the drive, as well as post-drive ratings of engagement, workload, and perceived safety for manual versus automated driving. Post-drive interviews were conducted to better understand participants' attitudes and their behaviours that were observed. Technology acceptance was higher following the drive as indicated by increased trust and perceived safety. Engagement was lower for automated driving, while no differences were reported for workload or safety between the driving modes. Perceptions of level-2 technology were highly positive, though complacent attitudes and behaviours were apparent. There was dissatisfaction with the hands-on steering wheel sensor, and observed behaviours indicate that it fails to adequately measure driver engagement. Many unintentional automation disengagements occurred due to problems interacting with the physical controls. Concerningly, this sometimes led to situations where the participant incorrectly thought they were still in automated mode, but were in fact responsible for primary driving tasks. Insufficient cues to inform the driver of mode changes and the current automation status contributed to the mode confusion observed. These findings can help inform the design of automated vehicle technology as well as policies and regulations, to ensure the inherent risks in partial driving automation can be managed.