Learning Low-Level Causal Relations using a Simulated Robotic Arm
Abstract
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.
Cite
@article{arxiv.2410.07751,
title = {Learning Low-Level Causal Relations using a Simulated Robotic Arm},
author = {Miroslav Cibula and Matthias Kerzel and Igor Farkaš},
journal= {arXiv preprint arXiv:2410.07751},
year = {2024}
}
Comments
14 pages, 5 figures, 3 tables. Appeared in 2024 International Conference on Artificial Neural Networks (ICANN) proceedings. Published version copyrighted by Springer. This work was funded by the Horizon Europe Twinning project TERAIS, G.A. number 101079338 and in part by the Slovak Grant Agency for Science (VEGA), project 1/0373/23. The code can be found at https://doi.org/10.5281/zenodo.14550231