RiSE-Drive: Risk-Sensitive Long-Horizon Candidate Evaluation for End-to-End Autonomous Driving

Abstract

End-to-end autonomous driving systems increasingly rely on generating multiple trajectory candidates and selecting one for execution. In this setting, a central challenge is no longer only candidate generation, but how to evaluate candidates according to their long-horizon consequences under uncertainty. Existing methods still mainly use deterministic point scores, heuristic subscores, or single-outcome evaluations, which makes it difficult to capture delayed failures, tail risk, and ranking instability under plausible future perturbations. We present RiSE-Drive, a framework for risk-sensitive long-horizon candidate evaluation in end-to-end autonomous driving. RiSE-Drive first performs candidate-conditioned latent future rollout, then estimates long-horizon outcome distributions instead of single scalar scores, and finally ranks candidates using a tail-risk-aware and perturbation-robust decision criterion. This reformulates candidate scoring from deterministic oracle-style evaluation into distributional outcome assessment with explicit risk control. As a result, the evaluator is encouraged to prefer candidates that are not only strong under nominal conditions, but also reliable under long-horizon rollout and safety-critical deviations. Experiments on the NAVSIM-v1 and NAVSIM-v2 benchmarks, as well as the closed-loop HUGSIM benchmark, show that RiSE-Drive achieves state-of-the-art performance and significantly improves the robustness of candidate selection.

arXiv comment11 pages, 3 figures, 6 tables, including supplementary material. Published in ACM Multimedia 2026 (MM '26)

References in corpus (16)

• InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models
• Emu3: Next-Token Prediction is All You Need
• VADv2: End-to-End Vectorized Autonomous Driving via Probabilistic Planning
• Hydra-MDP: End-to-end Multimodal Planning with Multi-target Hydra-Distillation
• DRAMA: An Efficient End-to-end Motion Planner for Autonomous Driving with Mamba
• DiffVLA: Vision-Language Guided Diffusion Planning for Autonomous Driving
• DriveVLA-W0: World Models Amplify Data Scaling Law in Autonomous Driving
• TransDiffuser: Diverse Trajectory Generation with Decorrelated Multi-modal Representation for End-to-end Autonomous Driving
• Enhancing End-to-End Autonomous Driving with Latent World Model
• RAP: 3D Rasterization Augmented End-to-End Planning
• Centaur: Robust End-to-End Autonomous Driving with Test-Time Training
• MindDrive: An All-in-One Framework Bridging World Models and Vision-Language Model for End-to-End Autonomous Driving
• Hydra-MDP++: Advancing End-to-End Driving via Expert-Guided Hydra-Distillation
• MeanFuser: Fast One-Step Multi-Modal Trajectory Generation and Adaptive Reconstruction via MeanFlow for End-to-End Autonomous Driving
• SafeGen: Goal-Conditioned Video Diffusion of Safety-Critical Scenarios for VLM-Based Autonomous Driving
• NSF-HRPT: Neural Semantic Field meets Hierarchical Risk Perception Tree for Safety-Critical Scenario Assessment