Infrastructure-based LiDAR can proactively detect intersection hazards at the edge without manual data labeling, using trajectory prediction and surrogate safety metrics to anticipate multi-agent conflicts in real-time.
PRISA is a roadside LiDAR system that monitors urban intersections in real-time to detect traffic conflicts before crashes happen. It uses privacy-preserving sensors, automatically learns from observed traffic patterns, and predicts collisions using two safety metrics—one for head-on crashes and one for crossing/pedestrian interactions—running fast enough to work on edge hardware.