用高维动作空间提升机器人世界模型的规划效率
Lifting Embodied World Models for Planning and Control
- 将高层指令（2D目标点）转为底层关节动作序列
- 比直接搜索低层动作降低3.8倍关节误差
- 适合需要快速规划的人形机器人场景
具身智能体的世界模型可基于代理执行的动作预测未来观测。对于复杂具身形态，动作空间维度高且难以定义：例如精确控制人形代理需指定每个关节的运动。这使得世界模型难以控制且规划成本高昂，因基于搜索的方法（如CEM）随动作维度增加而性能急剧下降。为此，我们训练了一个轻量级策略，将高层动作映射为一系列低层关节动作。将该策略与冻结的世界模型结合，形成可从单一高层动作预测未来观测序列的提升型世界模型。我们在人形具身系统中实现该框架，将高层动作空间定义为当前观测帧上标注的少量2D目标点，每个点指定一个末端关节（骨盆、头部、手部）的短期目标位置。这些目标点维度低、视觉可解释且易于手动设定或搜索。实验表明，该提升型世界模型显著优于直接在低层关节空间搜索（平均关节误差降低3.8倍），同时保持更低计算开销，并能泛化至策略未见过的环境。
原文摘要 · Abstract (English)
World models of embodied agents predict future observations conditioned on an action taken by the agent. For complex embodiments, action spaces are high-dimensional and difficult to specify: for example, precisely controlling a human agent requires specifying the motion of each joint. This makes the world model hard to control and expensive to plan with as search-based methods like CEM scale poorly with action dimensionality. To address this issue, we train a lightweight policy that maps high-level actions to sequences of low-level joint actions. Composing this policy with the frozen world model produces a lifted world model that predicts a sequence of future observations from a single high-level action. We instantiate this framework for a human-like embodiment, defining the high-level action space as a small set of 2D waypoints annotated on the current observation frame, each specifying a near-term goal position for a leaf joint (pelvis, head, hands). Waypoints are low-dimensional, visually interpretable, and easy to specify manually or to search over. We show that the lifted world model substantially outperforms searching directly in low-level joint space ($3.8\times$ lower mean joint error to the goal pose), while remaining more compute-efficient and generalizing to environments unseen by the policy.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。