Human Factors Simulator Study in Michigan Assessed Driver Behavior When Using Intersection Alert and Automatic Control Systems.
Michigan, United States
Connected and Automated Vehicle Based Intersection Maneuver Assist Systems (CAVIMAS) and Their Impact on Driver Behavior, Acceptance, and Safety
Summary Information
Vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) safety systems may potentially mitigate intersection crashes but need to consider human factors relating to content of information presented to drivers. Researchers in Michigan developed a conceptual driver assist system, called Connected and Automated Vehicle based Intersection Maneuver Assist System (CAVIMAS), to support assessment of driver behaviors related to in-vehicle system interfaces. Three concepts were implemented in a high-fidelity driving simulator:
- Baseline – the driver did not receive any in-vehicle driver assistance/warnings apart from navigational guidance.
- Driver Alerts – the driver was provided with alerts such as “brake” and “speed up” in response to potential conflicts with incoming or crossing vehicles at intersections.
- Alerts & Vehicle Control – in addition to driver alerts, an automated vehicle control system took braking or acceleration action to avoid potential collisions and provided drivers with indications such as “Braking” or “Speeding Up”.
Methodology
Using 24 licensed driver participants driving in a simulated city environment, CAVIMAS was used to analyze behavior in three common intersection-related crash types, left turn across path – opposite direction (LTAP-OD), left turn across path – lateral direction (LTAP-LD), and straight crossing path (SCP). To measure driver reactions and performance to the intersection maneuver systems, a four-camera remote eye tracker was installed in the simulator to measure head-pose, eye-blink, fixation locations, fixation duration, and pupil diameter. Participants also completed two surveys to gather perceptions of the helpfulness of the visualizations, accuracy of the warnings, driver confidence in turning at intersections and perceived workload using Task Load Index (TLX) measures. Within-subject comparisons were conducted to examine differences in driver behaviors between baseline and the alerts and vehicle control conditions.
Findings
- The eye tracker analysis results indicated reduced external visual scanning behavior among drivers while using assistive systems. The odds of looking right when turning left in the alerts and vehicle control scenario was 0.2 times the odds of scanning in the baseline scenario.
- Based on the survey results, 70 percent of the drivers considered the intersection warnings helpful, even though 98.8 percent of them were confident of their driving skills.
- Driver perception surveys that measured driver trust and acceptance found a slightly higher percentage of drivers (58 percent) indicating trust in the alerts-only concept as compared to the system with vehicle control (50 percent).
- Statistical tests of survey responses revealed that drivers had a higher preference for the recommendation to brake as compared to the acceleration recommendation. Similarly, participants had higher satisfaction with the automatic braking component compared to the automatic acceleration feature.
