Progress toward fully automated vehicles is happening rapidly, but many questions remain unanswered – particularly those related to human behavior.

“Some of the pressing issues are those related to human factors,” said Anuj K. Pradhan, part of the University of Michigan Transportation Research Institute’s Young Driver Behavior and Injury Prevention Group. “Until vehicles are fully automated, 100-percent of the time, humans will still play an active and important role in the driving loop. So for automation levels 2 and 3, the human factor is critical.”

According to the National Highway Transportation Safety Administration, a fully automated vehicle allowing unmanned operation is designated as level 4. In level 3, a vehicle essentially self-drives and controls all safety-critical functions, but certain situations might require transition back to driver control.

Level 2 automation represents combinations of multiple functions that enable a car to drive itself but require constant monitoring by the driver. The transitioning between automated and human control might be rapid.

These transitions may not be as simple as they sound.

“When you’re driving [normally], one is expected to possess a certain degree of situational awareness, especially with regard to the driving scene, traffic flow, and other driving related factors,” Pradhan explained. “But when a vehicle is in automated mode, the person inside may be reading, texting, or even napping. He or she may not be prepared to suddenly and safely take control of the vehicle when control is handed back from the automation.”

So the question for researchers is this: How can you safely and economically conduct empirical research to better understand the human-factor issues inherent in automated driving?

One approach is to perform vehicle testing in an off-roadway setting such as M City, the 32-acre testing facility located at the university’s north campus research complex. But since access to self-driving cars is very limited, even for research purposes, an alternate approach is to use high-fidelity driving simulators with automated-vehicle capabilities, such as the one at UMTRI.

The project has drawn seed funding from U-M’s Mobility Transformation Center (MTC) and secured an industrial partner, Realtime Technologies Inc.

Pradhan is working with programmer Christopher Atkins and lead electronics engineer Mark Gilbert to enhance UMTRI’s driving simulator to provide different levels of automated-driving capabilities. The software and hardware enhancements will provide for various customizations and experimental setups and designs. As a result, researchers can study human behavior in the context of automated driving with a high degree of validity.

When Pradhan pushes a small button on the steering wheel of UMTRI’s driving simulator, a voice announces, “Automated mode engaged.” The instrument cluster takes on a green background, and the car centers itself in the lane and cruises forward at a steady speed. The steering wheel turns by itself to keep the vehicle in the lane during curves and turns.

“The driving simulator, as is, is a very powerful and versatile tool that is used to study driver behavior," Pradhan said. "By adding self-driving capabilities, it allows us to significantly extend human-factors research on vehicle automation and to start examining a host of new questions that have been raised about vehicle automation in relation to the human driver or operator.”

Pradhan offered these examples of pertinent questions related to the transfer of vehicle control:

• How do people react when control of the vehicle is transferred back to them from automation during different circumstances? How does this differ if the driver is already monitoring the roadway versus engaged in something else?

• How do individual driver characteristics – such as age or experience – affect behaviors related to automation, including acceptance of technology and expectations of automation? Are alerts related to transfer of control best relayed by single or multiple modes (audio and seat vibrations) depending on the driver’s age or experience?

In a related project recently awarded by the MTC, Pradhan will examine these types of behavioral issues and gain a better understanding of the role that the driver’s state of attention plays during vehicle automation.

“It appears that when a human is in a self-driving car, it becomes important for the automation to also monitor the driver,” Pradhan noted. “That’s the next step.”