Research Papers

The Influence of Interface Design on Driver Behavior in Partially Automated Driving

Filename 1C-Lau-FP.pdf
Filesize 782 KB
Version 1
Date added July 9, 2018
Downloaded 7 times/fois
Category 2018 CARSP XXVIII Victoria
Tags Research and Evaluation, Session 1C
Author/Auteur Lau, Harbluk, Burns, El-Hage
Stream/Volet Research and Evaluation
Slidedeck Presentation

1C - Lau


Partially automated systems can control the vehicle longitudinally and laterally. However, these systems require drivers to actively monitor the environment and be readily available to act during take-over requests, situations where the automation does not work, or automation failures. To ensure that the driver makes the appropriate, safe, and timely response, the Human Machine Interface (HMI) should provide adequate information about system status, any system changes and, if necessary, what actions the driver should take. Some HMIs designed for partially automated systems provides only limited information about the current state of the system. For instance, some interfaces are unclear at communicating to the driver which function of the system is engaged and what is expected from the driver when system status changes. Furthermore, some icons used to display information on system status and changes is poorly located and are often small and obstructed. This poses a problem as information about automation status can be easily overlooked by the driver. The goal of this study was to investigate the impact of interface designs (Simple and Advanced) on driver behaviour in a partially automated driving system. Thirty-two participants drove a simulator configured as a partially automated driving system. They experienced either the Simple or Advanced Interface. The two interfaces were designed to differ in the presentation format and amount of information communicated to the driver. The Simple Interface was designed to resemble displays in some currently available partially automated vehicles. The Advanced Interface provided more elaborated and clearly presented information. Participants performed a continuous dot-counting game during the drive but were expected to take appropriate actions during scenarios when system limits were exceeded. Measures of driver behaviour included time to respond to the interface alert, disengage from the counting task, hand grip on wheel, and steer. Measures of driver response time were obtained from the simulator and video reduction. Participants provided ratings of usability, clarity and preference for the interface as well as their ratings of mental effort for its use. Data were analyzed for two scenarios of interest, a vehicle cut-in and construction zone. In both scenarios, initial response times were significantly faster for drivers using the Advanced Interface compared to those using the Simple Interface. Furthermore, the Advanced Interface helped drivers prepare for action significantly faster than the Simple Interface during the vehicle cut-in scenario. This was indicated by faster task disengagement from the dot-counting game and faster hands-on-wheel time. In some cases, drivers using the Simple Interface were unsuccessful at making safety maneuvers for both scenarios in the time allocated. N/A Vehicle interface design is already a concern and risks will likely increase with more complex automated driving systems. This study found improved driver performance with an elaborated interface where drivers were faster to take appropriate avoidance actions. HMIs designed for partially automated systems should be clear about system status, capture attention quickly and effectively, and, ideally, provide useful information to guide the driver to take appropriate actions. There is a need to develop driving guidelines for the design of these interfaces as well as metrics to assess their safety.