arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2609.35047cs.ROcs.AIcs.LG

EMPIRIC:基于实验驱动的残差世界模型学习用于机器人规划

EMPIRIC: Experiment-Driven Learning of Residual World Models for Robot Planning

  • Basis Research Institute(基础研究院)
  • University of Cambridge(剑桥大学)
  • Carnegie Mellon University(卡内基梅隆大学)
  • Fondazione Bruno Kessler(布鲁诺·凯斯勒基金会)
  • Massachusetts Institute of Technology(麻省理工学院)
  • The Alan Turing Institute(艾伦·图灵研究所)
  • Princeton University(普林斯顿大学)
  • Cornell University(康奈尔大学)
  • Harvard University(哈佛大学)

机构由 AI 辅助整理,请以论文原文为准。

Yichao Liang, Amber Li, Dat Nguyen, Emily Bunnapradist, Michelangelo Naim, Sreela Kodali, Matteo Merler, Bowen Li, Kiran Gopinathan, Yiyun Liu, Nikhil Pimpalkha… 展开作者

Yichao Liang, Amber Li, Dat Nguyen, Emily Bunnapradist, Michelangelo Naim, Sreela Kodali, Matteo Merler, Bowen Li, Kiran Gopinathan, Yiyun Liu, Nikhil Pimpalkhare, Joshua B. Tenenbaum, Adrian Weller, Zenna Tavares, Tom Silver, Kevin Ellis

AI总结:

提出EMPIRIC智能体,通过实验学习残差世界模型,扩展物理引擎以处理缺失机制,在模拟和实体机器人上以更少交互解决更多任务。

AI中文摘要:

机器人应该能够通过实验学习不熟悉物体的行为及相互作用方式,并利用这些知识进行规划。它无需从零开始:物理引擎提供了运动和接触的知识,但可能遗漏整个机制,例如胶水固化、水加热或风力。我们提出了EMPIRIC,一种学习残差世界模型的智能体:该模型是一个物理引擎,并附加了用于缺失机制的代码。学习到的程序可以引入新的力、约束和隐藏状态,贝叶斯推断从噪声观测中估计其参数和状态。由此产生的模型使智能体能够预测行动的结果,选择信息量大的实验,并在预测失败时修正其假设。在五个模拟领域中,EMPIRIC学习到可解释、可复用的模型,并以比所有三种基线更少的环境交互解决了更多任务。在实体机器人上,它学习风力作用和多米诺骨牌质量以解决操作任务。网站和代码:此https URL

英文摘要:

A robot should be able to learn through experiments how unfamiliar objects behave and interact, then plan with that knowledge. It need not start from scratch: physics engines supply knowledge of motion and contact, but can omit entire mechanisms, such as glue curing, water heating, or wind. We present EMPIRIC, an agent that learns a residual world model: a physics engine extended with code for the missing mechanisms. The learned programs can introduce new forces, constraints, and hidden state, and Bayesian inference estimates their parameters and states from noisy observations. The resulting model lets the agent predict the outcomes of actions, choose informative experiments, and revise its hypotheses when predictions fail. Across five simulated domains, EMPIRIC learns interpretable, reusable models, and solves more tasks with fewer environment interactions than all three baselines. On a physical robot, it learns wind forces and domino masses to solve a manipulation task. Website and code: https://yichao-liang.github.io/empiric

补充信息

↑