Streamlining Driver Research into Humans/Autonomous Vehicle Interactions
Published
A recent article in the New York Times highlighted “a growing number of crashes involving [Tesla’s] Autopilot that have fueled concerns about the technology’s shortcomings.” This is calling into question the speed with which automakers are developing and deploying new ADAS (advanced driver-assistance system) features. These concerns should be guiding driver research. But, to date, research into the complex interactions between autonomous vehicle/ADAS and human motorists has hardly scratched the surface.
On the one hand, these are clearly systems with great potential to increase public safety and vastly increase the independence of elders and others with limited mobility options. But as the Congressional Research Service noted in a 2021 report: “While these technologies hold many potentially transformative benefits, accidents involving them—such as Uber and Tesla crashes in 2018 caused in part by the drivers’ over-reliance on the automated technologies in the vehicles—demonstrate their potential safety challenges.”
Driver Research Isn’t Keeping Up with Autonomous Vehicle Adoption
When we first raised the issue of driver over-reliance on ADAS several years ago, Heather Stoner (General Manager for Realtime Technologies) noted that “the toothpaste is out of the tube.” Tesla already had a product on the road. Motorists were buying it and using it—despite demonstrably having no clear notion of what its limitations were or how they should interact with it.
2016 was the first year that an auto death was attributed to Tesla’s Autopilot ADAS. At that time, Tesla only had around 50,000 vehicles with Autopilot on the road. In the intervening years, that number has quadrupled to roughly 200,691. And the market has diversified, with many auto manufacturers adding new low-level ADAS features. Both Ford and GM are currently planning to roll out systems comparable to Tesla’s Autopilot.
Meanwhile, we’ve made little progress in understanding how humans interact with these technologies or how we can train new drivers to use them to increase public safety. We still see an unacceptably high number of accidents, concerning incidents, and near misses where it is unclear what’s gone wrong. Are there issues with the ADAS systems? Are drivers using them improperly?
Better Tools for Human Factors Research
“[Tesla is] still seeing automation failures, as well as over-trust in automation by their customers,” Stoner notes. “The real problem is that we don’t have a sufficient body of research into the human-vehicle interaction. If you want to really see what people are going to do once they’re in an autonomous vehicle, they need to feel like they’re actually in a vehicle.”
This is why RTI has developed SimCreator DX. This drag-and-drop tool allows researchers to quickly build the scenarios they need for driver research and dig deeper into the challenges of human-vehicle interaction: