SAFER 框架破解自动驾驶世界模型规划的探索-利用难题
SAFER: Navigating the Exploration-Exploitation Dilemma in World-Model-Based Planning for Autonomous Driving
这篇论文解决了自动驾驶世界模型的一个真问题:训练数据没覆盖的场景模型会瞎猜。它靠专家标注别人造成的故障来离线扩展覆盖,CARLA 和 MetaDrive 上的闭环成绩都是对比方法里最高的。做自动驾驶规划或闭环仿真的开发者值得把方法部分细看。
arXiv 论文提出 SAFER 框架,针对基于世界模型的自动驾驶规划中的探索-利用难题:模型在训练数据覆盖范围内难以提升,超出覆盖范围则预测误差会误导规划器。
在 CARLA Leaderboard 闭环评测中,SAFER 从特权鸟瞰视角取得 0.95 路线完成率和 0.
77 每步奖励,均为对比方法中最高;在 MetaDrive 中取得最高回合奖励,常规交通为 53.
4,安全关键场景为 49.0。
其技术路径分两阶段:离线阶段通过部分训练的智能体和脚本化可恢复扰动在仿真中制造故障场景,由专家对每个访问状态标注动作与价值,让规划器的探索空间在实际上路前就被覆盖;在线阶段用分块潜变量世界模型做实时后果预测,规划器以回报减去有界碰撞惩罚评估候选动作,并持续在偏离离线数据覆盖的驾驶区域精调模型。
相比从交互中试错学习的方法,SAFER 借他人之失获取故障经验而无需自己失败,这种离线失败学习加在线精调的组合对从事闭环自动驾驶评测的团队有直接参考价值。
SAFER: Navigating the Exploration-Exploitation Dilemma in World-Model-Based Planning for Autonomous Driving
An autonomous vehicle must foresee the consequences of its actions, especially in the rare, time-critical situations where crashes happen. A world model offers this foresight, but only within the coverage of its training data, which poses an exploration-exploitation dilemma. Within this coverage a planner can hardly improve on the data, and beyond it the model's errors mislead it. Logs of competent driving rarely contain the failures that would widen the coverage, and agents that learn from interaction gain them only by failing themselves, without being shown what to do instead. We propose SAFER, a framework that learns offline from failures that others make, judged by an expert, and keeps learning online from its own driving. Offline, partially trained agents and scripted, recoverable perturbations Stage failures in simulation, and an expert Annotates every visited state with its action and value, so that much of what the planner explores is covered before it drives. In planning, a chunked latent world model Foresees the consequences fast enough for real-time use, and a risk-aware sampling-based planner Evaluates candidate maneuvers by their return minus a bounded collision penalty. Online, a loop continuously Refines the model wherever this driving leaves the coverage of the offline data. In closed-loop evaluation against the compared methods, SAFER reaches the highest route completion and reward per step on the CARLA Leaderboard routes, 0.95 and 0.77 from a privileged bird's-eye view, and the highest episode reward in MetaDrive, 53.4 in regular and 49.0 in safety-critical traffic. Project page: https://ys-qu.github.io/safer-website.
本条情报归属 AITOP「论文」核心追踪栏目。涉及核心实体与技术议题包括:世界模型、自动驾驶、CARLA、强化学习,读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。
信源核验说明:本文原始线索采编自「arXiv cs.AI」(@Yansong Qu, Zilin Huang, Zihao Sheng, Sikai Chen, Samuel Labi),经 AITOP 多维评分管线结构化提炼。