WPT：通过在线世界模型蒸馏实现从世界模型到策略的迁移
WPT: World-to-Policy Transfer via Online World Model Distillation
对做视觉驱动决策的研究者，可复用其「世界模型在线蒸馏到轻量学生策略」的范式：用世界模型预测的未来动力学构造奖励信号训练教师策略，再蒸馏为学生策略，从而在保持实时推理的同时获得世界模型的规划收益。
原文依据 we propose policy distillation and world reward distillation to transfer the teacher's reasoning ability into a lightweight student policy
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近年来，世界模型取得了显著进展，其主要目标是刻画智能体动作与不断变化的环境之间的时空关联。然而，现有方法往往存在运行时紧耦合的问题，或依赖离线奖励信号，从而导致可观的推理开销，或阻碍端到端优化。为克服这些局限，我们提出 WPT，一种在端到端世界模型引导下实现在线蒸馏的「从世界模型到策略」训练范式。具体而言，我们设计了一个可训练的奖励模型，通过将候选轨迹与世界模型预测的未来动力学对齐，把世界知识注入教师策略。随后，我们提出策略蒸馏与世界奖励蒸馏，将教师的推理能力迁移到轻量级学生策略中，在保持实时可部署性的同时提升规划性能。在开环与闭环基准上的大量实验表明，WPT 以简单的策略架构取得了最优性能：开环碰撞率为 0.11，闭环驾驶得分为 79.23，在准确性与安全性上均优于基于世界模型的方法与模仿学习方法。此外，学生策略的推理速度最高提升 4.9 倍，同时保留了大部分性能收益。
英文原文摘要 1399 字符
Recent years have witnessed remarkable progress in world models, which primarily aim to capture the spatiotemporal correlations between an agent's actions and the evolving environment. However, existing approaches often suffer from tight runtime coupling or depend on offline reward signals, resulting in substantial inference overhead or hindering end-to-end optimization. To overcome these limitations, we introduce WPT, a World-to-Policy Transfer training paradigm that enables online distillation under the guidance of an end-to-end world model. Specifically, we develop a trainable reward model that infuses world knowledge into a teacher policy by aligning candidate trajectories with the future dynamics predicted by the world model. Subsequently, we propose policy distillation and world reward distillation to transfer the teacher's reasoning ability into a lightweight student policy, enhancing planning performance while preserving real-time deployability. Extensive experiments on both open-loop and closed-loop benchmarks show that WPT achieves state-of-the-art performance with a simple policy architecture: it attains a 0.11 collision rate (open-loop) and achieves a 79.23 driving score (closed-loop), surpassing both world-model-based and imitation-learning methods in accuracy and safety. Moreover, the student sustains up to 4.9xfaster inference, while retaining most of the gains.