Scenario2Vector -- Scenario Description Language Based Embeddings for Traffic Situations
Abstract:
The industry standard metric for measuring progress in autonomous driving has been the “miles per intervention” metric. This is nowhere near a sufficient metric and it does not allow for a fair comparison between the capabilities of two autonomous vehicles (AVs). We propose Scenario2Vector - a Scenario Description Language (SDL) based embedding for traffic situations that would allow for automatically searching for similar traffic situations. The SDL embedding distills a traffic situation experienced by an AV into its canonical components - actors, actions, and the traffic scene. We can then use this embedding to evaluate similarity of different traffic situations in vector space.
Committee:
- Aidong Zhang, Committee Chair, (CS/SEAS/UVA)
- Madhur Behl, Advisor, (CS/SEAS/UVA)
- Jack Stankovic (CS/SEAS/UVA)
- Vicente Ordóñez Román (CS/SEAS/UVA)