Waymo has created a new computer model designed to more precisely answer a key question: How does its autonomous driving software stack up against humans?
Alphabet-owned robotaxi, which developed a computer model of human driving abilities in collaboration with TU Delft, published a study. paper About it Wednesday in Nature Communications.
Waymo said it expects the new model to be more accurate than the previous version it has been using for the past few years. The new model is built using a framework called active inference – the theory that the driver is constantly imagining possible futures and taking action to reach the safest, most predictable one.
Waymo said the new model will help better understand how humans behave in crash scenarios encountered by robotaxis.
“For decades, the auto industry has used physical and virtual crash dummies to evaluate a vehicle’s safety features, including its hardware and structural integrity,” Waymo wrote in a blog post Wednesday. Waymo said the new model “improves on this concept, serving as behavioral benchmarks for autonomous driving systems that can realistically reflect reasonable expectations of how a cautious and skilled human driver would react to traffic conflicts.”
A more accurate model of human driving behavior is table stakes for autonomous car companies that need to understand and evaluate the performance of their robotaxis in the event of accidents. And it comes at a critical time for Waymo, which is expanding into more cities and facing more scrutiny from regulators and the public.
In January a Waymo robotaxi hit the child Near a school in Santa Monica, Calif., the company cited an earlier computer model to claim that a careful human driver could make the impact at about 14 miles per hour. The Waymo robotaxi hit the boy at just 6 miles per hour after slowing down from 17 miles per hour, and the company said he suffered minor injuries. (The accident is still under investigation by the National Highway Traffic Safety Administration and the National Transportation Safety Board.)
The biggest difference between this new model, which Waymo calls the Reference Driver, and its predecessor is that it can mimic the behavior of a human driver in the run-up to a crash. Previously, Waymo’s models (and other industry models) focused on replicating “last-second, reactive” human maneuvers, according to the company.
Meanwhile, the Reference Driver “can simulate the internal ‘surprise’ a driver feels during a conflict and provide a more human-like benchmark for autonomous driving systems that were previously impossible to automate at scale,” said TU Delft associate professor Arkady Zgonnikov.
Waymo says this new driver model can be adapted to model “a wide range of road user behaviors beyond collision avoidance” and is better equipped to apply to its “large test sets with thousands of scenarios.”
“The model can represent and evaluate multiple complex, real-world accidents in a virtual environment and identify performance improvements with unprecedented speed and efficiency,” the company wrote.
Waymo wants others to collaborate to take Reference Driver even further. The company said Wednesday that it is making the research code for the model available under an academic, non-commercial license, which allows it to be used for research, teaching, personal practice and scientific publication.
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
