Brief

Researchers have developed RECTOR, a novel reranking system designed to improve the safety and compliance of autonomous driving trajectory selections. This system prioritizes safety, legal adherence, and comfort rules over simple model confidence scores. By employing a tiered rulebook and a differentiable proxy mechanism, RECTOR significantly reduces violations compared to confidence-only methods, even under adversarial conditions. AI

Researchers have developed RLFTSim, a new framework for creating more realistic and controllable multi-agent traffic simulations. This system uses reinforcement learning to fine-tune existing simulation models, aligning their outputs with real-world driving data distributions. Experiments on the Waymo Open Motion Dataset show RLFTSim achieves state-of-the-art realism and requires fewer samples than other methods due to its reward design. AI