Waymo has developed a new computer model designed to simulate human driver behavior in challenging road scenarios, the autonomous vehicle company announced. The virtual driver system is intended to help the company better understand how humans react to unexpected situations and crashes that its robotaxis encounter in real-world conditions.

The model represents what Waymo describes as a "hyperattentive driver" that can be tested against the company's autonomous vehicles in simulated scenarios. This virtual representation allows researchers to study human responses to surprises and unpredictable situations on the road without putting real drivers at risk.

The development builds on Waymo's existing simulation capabilities, which the company has used extensively to train and test its autonomous driving systems. Previously, Waymo created realistic 3D virtual worlds to help its vehicles better anticipate natural disasters and other edge cases that might not occur frequently in real-world testing.

The new benchmark tool is designed to provide more accurate comparisons between autonomous vehicle performance and human driver behavior in crash scenarios. By creating a standardized virtual human driver, Waymo aims to establish clearer metrics for evaluating how its robotaxis perform relative to human drivers in similar situations.

Waymo's simulation approach reflects the broader industry trend of using virtual testing environments to supplement real-world autonomous vehicle development. These computer models allow companies to test millions of scenarios that would be impractical or dangerous to recreate with actual vehicles and human drivers on public roads.