Different HMI Configurations for Automated Vehicles Tested with 54 Participants in Amherst, Massachusetts.
Amherst, Massachusetts, United States
Human-Machine Interfaces and Vehicle Automation: The Effect of HMI Design on Driver Performance and Behavior
Summary Information
Recent vehicle automation technologies offer great potential to support drivers and reduce human errors that result in crashes. However, the effectiveness of a Human–Machine Interface (HMI) in an automated driving system during takeover situations is often based on its design in terms of modality, specificity, and timing of the HMI. This study examined the effectiveness of two HMIs, varying in their modality, specificity, and timing, on drivers’ takeover time, performance, eye glance behavior, and subjective evaluation. Driver behaviors were examined in a driving simulator study with 54 participants recruited from the University of Massachusetts–Amherst campus. The study assessed different levels of automation, traffic conditions, and the impact of multitasking with a non-driving related task.
METHODOLOGY
A driving simulator study was conducted in which two groups of drivers experienced one of two different HMI designs. One group experienced a Staged HMI design including a two-stage multimodal alert system, in which the drivers received a visual warning message followed by a non-descriptive auditory beep about five seconds later. The second group experienced a Simultaneous HMI design including a one-stage multimodal alert system where the drivers received a visual and auditory (explanatory) message at the same time. For each group of drivers, the HMI was implemented at four different Society of Automotive Engineers (SAE) levels of automation (Level Zero to Three), with each involving situations such as weather, traffic, or road conditions on a straight roadway with a speed limit of 65 miles per hour. To simulate real-world behaviors, non-driving related tasks were included during Level Three automation drives.
FINDINGS
Results indicated that HMI type had an effect on velocity and off-road eye glances.
- For Level One automation, results showed that vehicle speeds were reduced by up to three percent after system takeover; For Level Two automation, results saw up to two percent increase (disbenefit) in speeds; For Level Three automation, speeds were reduced by up to 16 percent following system takeover of driving.
- Comparing Staged and Simultaneous HMI, for Level One automation, the total off-road glance duration was reduced by 18.8 percent. For Level Three automation, the reduction was 37.2 percent. However, there was an 18.6 percent increase (disbenefit) for Level Two automation.
- In terms of the frequency of off-road glances, results indicated improvement across all levels of automation. Specifically, for Level One, the reduction in the frequency of glances was 20.6 percent, for Level Two it was 25.5 percent, and for Level Three automation the reduction was 18.8 percent.
