Waymo Publishes New Computer Model for Comparing Robotaxi and Human Driving Behavior
Waymo released a research paper describing a new computer model of human driving behavior. The model was developed with TU Delft and published in Nature Communications. It is intended to provide a more detailed benchmark for evaluating robotaxi performance in crash scenarios.
TechCrunchWaymo published a research paper on Wednesday describing a new computer model of human driving behavior. The Alphabet-owned company developed the model with TU Delft and released the findings in Nature Communications. The model uses a framework called active inference to simulate how drivers anticipate possible outcomes and select actions.
Waymo stated that the model is intended to be more accurate than the version it has used in recent years.
The company said the model can reproduce a human driver's behavior in the period leading up to a crash, rather than only last-second reactions. It can also be applied to large sets of scenarios, according to the paper. Waymo wrote that the model serves as a behavioral benchmark for autonomous systems and can represent expectations for how a careful human driver would respond to traffic conflicts.
The company said it is releasing the research code under an academic, non-commercial license for use in research, teaching, and scientific publication. It also said the model can be adapted to represent a range of road user behaviors beyond collision avoidance.
In January, the company used its previous model to compare its robotaxi's speed to an estimated human driver speed in a Santa Monica incident that remains under review by federal safety agencies.
