News | CSULB's Researchers James Miles and Thomas Strybel Evaluate the Safety of Autonomous Vehicles

Stop the Video

News

METRANS

Associate Professor James Miles

     

Assoc. Professor Jim Miles            Professor Emeritus Thomas Strybel

 

Automation continues to impact transportation, compelling the industry to consider how to implement transformative mobility developments in a safe and efficient manner. While fully autonomous vehicles, vehicles having the ability to safely reach a destination without human intervention, are highly anticipated, current technology allows only partial automation, and requires an engaged driver’s input for safe operation. The need for drivers to maintain vigilance and be prepared to take over driving with partial automation systems raises the question of their safety in comparison to manual driving. At California State University, Long Beach, Psychology Associate Professor James Miles and Professor Emeritus Thomas Strybel studied the safety of highly automated driving systems (HADS) for human- driver takeover performance in comparison with conventional manual driving in their PSR-funded project, “Evaluation of Autonomous Vehicles and Smart Technologies for Their Impact on Traffic Safety and Traffic Congestion.”

Miles and Strybel focused on “relations between measures of workload, opinions of automation, and driver performance...to determine whether these factors predict successful performance on driver events (obstacles and curves) in both the manual and take-over." The Society of Automotive Engineers (SAE) laid out six levels of automation ranging from Level 0 where drivers maintain full vehicular control even with driver assist features and Level 5 where automation controls all driver functions even with a “driver” in the seat. Miles and Strybel concentrated specifically on the levels that involve automated driving systems that would not require constant human driver intervention. Noting research gaps in comparisons of manual driving and the takeover performance of HADS, the study used a virtual-reality simulation of eight test tracks in their work, allowing thirty-seven participants to drive the courses both manually and using HADS.

The study found that emergency avoidance maneuvers were performed more safely following HADS takeovers than during traditional all-manual driving. The authors speculate that most take-over requests (TORs) from HADS vehicles will likely occur immediately before an emergency driving maneuver is needed, emphasizing that TORs may serve a dual function by indicating the need to resume control of the vehicle while alerting the driver of an impending road hazard. The researchers further concluded that HADS conditions promoted safe operations, such as driving at lower speeds and “less lane deviation,” thus leading to safer driving performance following HADS take-overs than fully manual driving under certain circumstances. HADS may also reduce general driver fatigue, allowing drivers to stay alert and prepared for take-over situations.

 

While the study had some limitations (e.g. the lack of measurement of the cultural and technological backgrounds of participants and the use of a relatively small selection of road hazards), the authors concluded that it should not be assumed that HADS takeovers will inevitably lead to worse driver performance in all situations. Automated systems like HADS in highway vehicles demonstrate that a bright future of safe and freely flowing roads may await us.