Realtime Technologies


Exploring Driver Disengagement During AV/ADAS Operation

It’s become abundantly clear that, as vehicles themselves become more capable of taking on routine driving tasks, human factors research has become even more critical for roadway safety. An increasing number of “autonomous” and ADAS-Enhanced (advanced driver-assistance system) vehicles are entering the market and joining us on roadways. Unfortunately, surveys have found that roughly half of all drivers fundamentally misunderstand what current “autonomous” and ADAS-equipped vehicles are capable of. For example, 48% of those surveyed believed it was safe to take your hands off the wheel when the ADAS is engaged. (It’s not.  Tesla’s owner’s manual even refers to their “Autopilot” steering function as a “hands-on feature”.) Up to 6% think it’s OK to take a nap(!!!) while using some of the ADAS features currently on the market. (It isn’t.)

As Helen S. Loeb (a researcher with the Children’s Hospital of Philadelphia) wrote earlier this year when discussing her most recent human factors research,  “vehicles are now much more than the integration of automotive elements; they are literally computers on wheels for which Human Machine Interfaces must carefully be designed.” Loeb recently co-authored a paper (“Vehicle Automation Emergency Scenario: Using a Driving Simulator to Assess the Impact of Hand and Foot Placement on Reaction Time“) which helps establish an important baseline for how average drivers behave behind the wheel of an “autonomous” vehicle.

Loeb notes that “[a]s vehicles with SAE level 2 of autonomy become more widely deployed, they still rely on the human driver to monitor the driving task and take control during emergencies. It is therefore necessary to examine the Human Factors affecting a driver’s ability to recognize and execute a steering or pedal action in response to a dangerous situation when the autonomous system abruptly requests human intervention.”


Using Real World Events in Human Factors Research

For Loeb’s human factors research, a group of drivers across all age groups were introduced to an immersive vehicle simulator from Realtime Technologies running a program emulating a vehicle with SAE Level 2 features similar to those on the market today. The simulated vehicle has the ability to maintain a specified driving speed and stay in the lane on its own. The participants received instructions on how to operate this vehicle’s ADAS features, and then took several 10 to 15 minute drives. What they did not know was that this scripted scenario included an emergency. At some point during their drive the automation would fail (in a manner based on documented real life events): while navigating a curve at around 30mph, the simulated ADAS-enhanced vehicle fails to properly track the lane markers and steers directly into oncoming traffic. Without the driver’s intervention, they will crash head-on.

Given that the average human reaction time behind the wheel is around .75 to 1 second, and the programmed “time to crash” in the scenario was 2.2 seconds, this should have been a close call that most drivers would be able to manage (if barely).


Human Factors Research Helps Us Understand Drivers in the Real World

How did they do?  Most drivers crashed when “assisted” with these autonomous safety features. In the experimental group more than half the drivers (53%) crashed. Almost all of the older drivers (65+) crashed, with 44% unable to react at all (compared to just 6% of adult and younger drivers).

Importantly, during the study—and immediately before their drive—participants were explicitly told about the limitations of SAE Level 2 ADAS. They were reminded that these systems rely on being able to fallback to full human control at a moment’s notice, and explicitly told that they were responsible for driving safely and avoiding crashes. Nonetheless, Loeb and her team found that despite explicit warnings about the limitations of the simulated ADAS system, 88% of drivers in the study had their hands off the wheel when the scripted emergency triggered. Almost half didn’t even have their feet near the pedals. In fact, around 7% had their foot under the brake pedal!

This is just one study in a growing body of human factors research that seems to indicate that simple education about the limitations and appropriate uses of ADAS and autonomous vehicle features just won’t be sufficient. Automakers likely need to explore designs that will help the vehicle know when drivers are disengaging and help coax them back to attending to their responsibilities behind the wheel.