arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2410.07751cs.ROcs.AIcs.LG
收藏

使用模拟机械臂学习低层因果关系

Learning Low-Level Causal Relations using a Simulated Robotic Arm

Comenius University Bratislava(布拉迪斯拉发考门斯基大学);Faculty of Mathematics, Physics and Informatics, Comenius University Bratislava(布拉迪斯拉发考门斯基大学数学、物理与信息学院);University of Hamburg(汉堡大学)

Miroslav Cibula, Matthias Kerzel, Igor Farkaš

AI总结:

本文利用模拟机械臂的感觉运动数据学习前向与逆向模型,并通过特征归因分析揭示低层因果关系,为状态降维和高层因果可解释性提供基础。

摘要:

因果学习使人类能够预测自身行为对已知环境的影响,并利用这一知识规划更复杂动作的执行。此类知识还刻画了环境的行为,可用于环境分析以及行为背后的推理。这种知识在具备常识的智能机器人系统设计中也至关重要。本文通过基于一个模拟机械臂在两项感觉运动任务中生成的数据,学习前向模型和逆向模型来研究因果关系。下一步,我们研究用于分析前向模型的特征归因方法,该方法揭示了与机械臂关节和环境特征相关的状态向量中各个特征所对应的低层因果效应。这种分析为状态表示的降维以及知识聚合以实现更高层次因果效应的可解释性提供了坚实基础。

英文摘要:

Causal learning allows humans to predict the effect of their actions on the known environment and use this knowledge to plan the execution of more complex actions. Such knowledge also captures the behaviour of the environment and can be used for its analysis and the reasoning behind the behaviour. This type of knowledge is also crucial in the design of intelligent robotic systems with common sense. In this paper, we study causal relations by learning the forward and inverse models based on data generated by a simulated robotic arm involved in two sensorimotor tasks. As a next step, we investigate feature attribution methods for the analysis of the forward model, which reveals the low-level causal effects corresponding to individual features of the state vector related to both the arm joints and the environment features. This type of analysis provides solid ground for dimensionality reduction of the state representations, as well as for the aggregation of knowledge towards the explainability of causal effects at higher levels.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑