Waymo Unveils 'Reference Driver' AI to Benchmark Autonomous Vehicle Safety
Waymo, Alphabet's self-driving unit, has developed a new AI model called 'Reference Driver' (ReD) in collaboration with TU Delft. Published in Nature Communications, the model simulates human driving behavior to benchmark autonomous vehicle safety and crash avoidance against human drivers, aiming for industry-wide safety standard development.
Key points
- Waymo, Alphabet's autonomous vehicle company, developed a cognitive model named 'Reference Driver' (ReD).
- The model simulates how competent human drivers make split-second decisions in traffic conflicts.
- It was created in partnership with the Delft University of Technology and uses active inference principles.
- Waymo published research on ReD in the journal Nature Communications, positioning it as a behavioral benchmark.
- The company aims to use ReD to compare its autonomous driving systems against human performance and advance industry safety standards.
Waymo, the autonomous driving technology division of Alphabet, has introduced a sophisticated new AI model designed to accurately simulate human driving behavior. Dubbed the 'Reference Driver' (ReD), this cognitive model was developed in collaboration with the Delft University of Technology in the Netherlands.
The research, published in the scientific journal Nature Communications, details how ReD operates by employing a framework known as active inference. This theory posits that drivers constantly anticipate future scenarios and choose actions to achieve the safest and most predictable outcomes. Waymo intends for ReD to serve as a critical tool for evaluating the safety performance of its autonomous vehicles, particularly in challenging crash-avoidance situations.
This development represents a significant step in Waymo's ongoing efforts to establish robust safety benchmarks within the autonomous vehicle industry. By creating a "behavioral dummy" analogous to physical crash test dummies used in automotive safety, Waymo aims to provide a standardized method for comparing robotaxi capabilities against realistic human driving responses. The company believes this will foster greater collaboration and help move the industry toward adopting more unified safety standards.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.