World4Scorer: Outcome-Grounded World Modeling for Autonomous Driving
作者: Jieyuan Pei, Meiyi Lu, Sining Ang, Yubo Zhao, Zhangyi Hu, Mingwei Xu, Haokai Ding, Wei Li, Zihan You, Jianwei Zheng, Li Yu, Yifeng Pan, Ji Tao, Rongjunchen Zhang, Yan Wang
分类: cs.RO
发布日期: 2026-09-29
备注: 28 pages, 14 figures, 13 tables. Project page: https://guobapei.github.io/World4Scorer/
💡 一句话要点
提出World4Scorer以解决自主驾驶中的轨迹评分问题
🎯 匹配领域: 支柱一：机器人控制 (Robot Control) 支柱二：RL算法与架构 (RL & Architecture)
关键词: 自主驾驶 轨迹规划 评分系统 机器学习 模拟器 决策支持 智能交通
📋 核心要点
- 现有的自主驾驶轨迹评分方法在处理未执行计划时缺乏有效的约束，导致评分不准确。
- 本文提出World4Scorer，通过构建轨迹条件的JEPA风格预测器，利用模拟器结果标签来监督所有候选轨迹的状态。
- World4Scorer在NAVSIM-v2上达到了最先进的性能，并在OGBench-Cube基准测试中提升了操作规划的效果。
📝 摘要（中文）
自主驾驶需要在周围交通不断变化的情况下选择安全高效的行驶计划。生成与选择规划器提出多条轨迹并对其进行评分，已在NAVSIM上超越了直接预测基线。现有方法的评分器必须比较未执行的计划，而驾驶日志仅记录已执行轨迹的未来，导致对未执行计划的预测缺乏约束。为此，本文提出World4Scorer，构建了一个基于轨迹条件的JEPA风格预测器，能够为每个候选轨迹预测状态，并从该状态中读取候选轨迹的评分。通过模拟器的结果标签来监督所有候选的状态，并通过已执行轨迹的观察未来来锚定预测器。实验结果表明，World4Scorer在NAVSIM-v2上实现了最先进的性能，并在闭环Bench2Drive上取得了强劲的适应系统结果。
🔬 方法详解
问题定义：本文旨在解决自主驾驶中轨迹评分的准确性问题。现有方法在比较未执行计划时，依赖于已执行轨迹的未来信息，导致评分缺乏约束，影响决策质量。
核心思路：提出World4Scorer，通过构建一个轨迹条件的JEPA风格预测器，能够为每个候选轨迹预测状态，并从中读取评分。利用模拟器的结果标签来监督所有候选轨迹的状态，从而增强评分的准确性。
技术框架：整体架构包括轨迹生成、状态预测和评分模块。首先生成多个候选轨迹，然后通过预测器为每个轨迹预测状态，最后根据状态输出评分。
关键创新：World4Scorer的主要创新在于通过一个预测器为所有候选轨迹生成状态，从而实现共享参数的约束，显著提高了未执行计划的评分准确性。
关键设计：在设计中，使用了模拟器结果标签作为监督信号，并在训练阶段依赖于已执行轨迹的观察未来。此外，采用了场景匹配的银行来增加低评分计划的监督，确保评分的一致性。
🖼️ 关键图片
📊 实验亮点
World4Scorer在NAVSIM-v2上实现了最先进的性能，相较于传统方法，提升了轨迹评分的准确性。此外，在闭环Bench2Drive上也取得了显著的适应系统结果，展示了其在实际应用中的有效性。
🎯 应用场景
该研究的潜在应用领域包括自动驾驶系统的轨迹规划与决策支持，能够提高自主驾驶车辆在复杂交通环境中的安全性和效率。未来，World4Scorer可能在其他需要实时决策的机器人系统中得到应用，推动智能交通的发展。
📄 摘要（原文）
Autonomous driving requires choosing a safe and efficient plan as surrounding traffic evolves. Generate-and-select planners propose multiple trajectories and score them for execution, and they have outperformed representative direct-prediction baselines on NAVSIM. Their scorer must compare plans that were never executed. Driving logs record the future of only the executed trajectory, so matching the logged future can leave predictions for the alternatives unconstrained; a simulator, in contrast, can label the outcome of every candidate. We introduce World4Scorer, which builds the scorer as a trajectory-conditioned JEPA-style predictor: it predicts a state for each candidate and reads the candidate's scores from that state. Simulator outcome labels supervise the states of all candidates, and the observed future of the executed trajectory anchors the predictor to real scene evolution. Because one predictor produces every candidate's state, the anchor can constrain shared parameters used to score unexecuted plans, while the future itself is needed only during training. Generated candidates mostly score well, so a scene-matched bank adds low-scoring plans to the outcome supervision; framewise choices can conflict, so inertial re-ranking keeps consecutive selections consistent. World4Scorer achieves state-of-the-art NAVSIM-v2 performance and a strong adapted-system result on closed-loop Bench2Drive. With the LeWM world model and planning budget fixed, outcome-based scoring also improves manipulation planning on the OGBench-Cube benchmark.