New driving planner improves safety and maintains diverse routes
Beyond Waypoint Regression: Query-Based Cost Learning over Reachable Ego Futures for End-to-End Driving
RoboticsArtificial IntelligenceMachine Learning
Summary
Driving systems often learn to follow smooth paths but struggle to consider safety rules when making decisions. The authors developed a new method that predicts costs for potential future routes the car could take, helping it pick safer and more practical paths. Their approach uses compact representations of traffic situations and balances different options to keep the driving plans both safe and varied. Tests on real-world driving data and simulations show fewer collisions compared to previous methods, even without extra training.
What this means in practice
- •For autonomous vehicle engineers: Create driving planners that better balance safety constraints and diverse route options without extra tuning on new data.$Commercial implications: Enables safer, commercially viable autonomous driving software that adapts to complex environments and reduces collisions using learned cost topologies.
- •For robotics control teams: Integrate cost-based trajectory evaluation strategies to improve real-time decision making in mobile robots navigating dynamic environments.
Authors
Ahmed Abouelazm, Rupert Polley, Qingyuan Zhang, Yin Wu, Philip Schörner, Carl Esselborn, J. Marius Zöllner
Abstract
End-to-end planners based on waypoint regression achieve strong open-loop accuracy, but they primarily learn to mimic expert geometry and remain difficult to adapt to deployment-time safety constraints. We propose a query-based cost-learning framework that estimates bounded costs for dynamically reachable ego trajectory queries, rather than dense BEV cells or a small regressed trajectory set. Compact joint scene tokens capture coherent multimodal agent futures, while contingency-aware cost aggregation and cost-guided intra-cluster MPPI mixing convert the learned cost topology into feasible ego plans. On nuScenes, our method improves over prior cost-estimation planners such as ST-P3 and NMP, outperforms most regression baselines in collision rate, while remaining competitive in L2, and retaining an interpretable cost interface. On real-world driving logs, the proposed planner reduces collision rates compared with SparseDrive and Alpamayo without fine-tuning, while maintaining a diverse set of candidate trajectories.