Drive in Corridors: Enhancing the Safety of End-To-End Autonomous Driving Via Corridor Learning and Planning
Zhiwei Zhang, Ruichen Yang, Ke Wu, Zijun Xu, Jingchu Liu, Lisen Mu, Zhongxue Gan, Wenchao Ding
AI summary
Problem
End-to-end autonomous driving models lack explicit safety constraints and mathematical guarantees, making them prone to collisions and difficult to interpret in dynamic traffic.
Approach
The method learns a spatio-temporal safe corridor from camera inputs and uses it as a differentiable constraint in a trajectory optimization process to generate kinematically feasible, collision-free paths.
Key results
- 66.7% reduction in agent collisions on nuScenes
- 46.5% reduction in curb collisions on nuScenes
- Improved closed-loop success rates on Bench2Drive
- First differentiable optimization layer integrated into end-to-end driving
Why it matters
It bridges learning-based planning and robotics safety guarantees, offering a scalable path to deploy interpretable and collision-avoidant autonomous vehicles.
Abstract
Safety remains one of the most critical challenges in autonomous driving systems. In recent years, the end-to-end driving has shown great promise in advancing vehicle autonomy in a scalable manner. However, existing approaches often face safety risks due to the lack of explicit behavior constraints. To address this issue, we uncover a new paradigm by introducing the corridor as the intermediate representation. Widely adopted in robotics planning, the corridors represents spatio-temporal obstacle-free zones for the vehicle to traverse. To ensure accurate corridor prediction in diverse traffic scenarios, we develop a comprehensive learning pipeline including data annotation, archi- tecture refinement and loss formulation. The predicted corridor is further integrated as the constraint in a trajectory optimization process. By extending the differentiability of the optimization, we enable the optimized trajectory to be seamlessly trained within the end-to-end learning framework, improving both safety and interpretability. Experimental results on the nuScenes dataset demonstrate state-of-the-art performance of our approach, show- ing a 66.7% reduction in collisions with agents and a 46.5% reduction with curbs, significantly enhancing the safety of end-to- end driving. Additionally, incorporating the corridor contributes to higher success rates in closed-loop evaluations. Project page: https://zhiwei-pg.github.io/Drive-in-Corridors.