Vision language model

AI system called KEPT lets self-driving cars use scene memory to predict safer paths and reduce planning errors.