SAH-Drive: A Scenario-Aware Hybrid Planner for Closed-Loop Vehicle Trajectory Generation

Reliable planning is crucial for achieving autonomous driving. Rule-based planners are efficient but lack generalization, while learning-based planners excel in generalization yet have limitations in real-time performance and interpretability. In long-tail scenarios, these challenges make planning particularly difficult. To leverage the strengths of both rule-based and learning-based planners, we proposed the Scenario-Aware Hybrid Planner (SAH-Drive) for closed-loop vehicle trajectory planning. Inspired by human driving behavior, SAH-Drive combines a lightweight rule-based planner and a comprehensive learning-based planner, utilizing a dual-timescale decision neuron to determine the final trajectory. To enhance the computational efficiency and robustness of the hybrid planner, we also employed a diffusion proposal number regulator and a trajectory fusion module. The experimental results show that the proposed method significantly improves the generalization capability of the planning system, achieving state-of-the-art performance in interPlan, while maintaining computational efficiency without incurring substantial additional runtime.

Autonomous vehicles need to plan safe and efficient routes in real-time. Traditionally, engineers have used rule-based systems for this, like giving the car a detailed list of "if-then" instructions. These are fast and predictable but don’t handle unfamiliar or complex situations well. On the other hand, AI-based planners can adapt to new scenarios by learning from data, but they can be slow or hard to understand.To get the best of both worlds, we created a new system called SAH-Drive. It works a bit like how human drivers switch between habit and conscious thinking: using simple rules for routine driving and a more powerful AI when the situation gets tricky. The system includes a smart decision-making unit that decides when to rely on which part. Our approach outperforms existing methods in tough driving scenarios while still running efficiently — a step closer to making self-driving cars safer and more dependable in the real world.