Driving, Fast or Slow? Neuro-Symbolic Guidance for Motion Prediction in Multi-Modal Ground Mobility

The article introduces Trajectory Compliance-Shaping (TraCS), a neuro-symbolic framework designed to enhance motion prediction in complex traffic environments by integrating first-order logic with existing neural models. TraCS employs an agentic code-generation pipeline for interpreting traffic regulations and features a reactive data-streaming inference engine for dynamic scene updates. Evaluated on the Argoverse 2 benchmark, TraCS shows significant improvements over traditional black-box models by incorporating probabilistic compliance reasoning, which is crucial for developing interpretable and safe autonomous navigation systems.