让机器人学会抽象世界模型，应对环境自发变化的挑战
ExoPredicator: Learning Abstract Models of Dynamic Worlds for Robot Planning
- 用变分贝叶斯+大模型提议，联合学习符号状态与因果过程
- 在5个模拟环境中实现快速规划，新任务泛化性能优于基线
- 适合需要长时序推理的机器人场景，如复杂物理交互
长时序具身规划面临挑战，因世界不仅受智能体动作影响，还存在外部过程（如水加热、多米诺骨牌连锁反应）同时演进。本文提出一种抽象世界建模框架，联合学习符号状态表示和内生动作与外生机制的因果过程。每个因果过程建模随机因果关系的时间演化。通过变分贝叶斯推断结合大语言模型提议，从有限数据中学习这些模型。在五个模拟桌面上机器人环境测试中，所学模型支持快速规划，能泛化至包含更多物体和更复杂目标的未见任务，性能超越多种基线方法。
原文摘要 · Abstract (English)
Long-horizon embodied planning is challenging because the world does not only change through an agent's actions: exogenous processes (e.g., water heating, dominoes cascading) unfold concurrently with the agent's actions. We propose a framework for abstract world models that jointly learns (i) symbolic state representations and (ii) causal processes for both endogenous actions and exogenous mechanisms. Each causal process models the time course of a stochastic cause-effect relation. We learn these world models from limited data via variational Bayesian inference combined with LLM proposals. Across five simulated tabletop robotics environments, the learned models enable fast planning that generalizes to held-out tasks with more objects and more complex goals, outperforming a range of baselines.
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。