Learning Causal Structure Distributions for Robust Planning

2025/08/08 by Alejandro Murillo-González, Junhong Xu, Murillo-Gonzalez, Alejandro +3
Computer Science · #AI-based Problem Solving and Planning #Artificial Intelligence (cs.AI) #Bayesian Modeling and Causal Inference #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Reinforcement Learning in Robotics #Robotics (cs.RO) #Systems and Control (eess.SY) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2508.06742

openalex publication_date 2025/08/08 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Structural causal models describe how the components of a robotic system interact. They provide both structural and functional information about the relationships that are present in the system. The structural information outlines the variables among which there is interaction. The functional information describes how such interactions work, via equations or learned models. In this paper we find that learning the functional relationships while accounting for the uncertainty about the structural information leads to more robust dynamics models which improves downstream planning, while using significantly lower computational resources. This in contrast with common model-learning methods that ignore the causal structure and fail to leverage the sparsity of interactions in robotic systems. We achieve this by estimating a causal structure distribution that is used to sample causal graphs that inform the latent-space representations in an encoder-multidecoder probabilistic model. We show that our model can be used to learn the dynamics of a robot, which together with a sampling-based planner can be used to perform new tasks in novel environments, provided an objective function for the new requirement is available. We validate our method using manipulators and mobile robots in both simulation and the real-world. Additionally, we validate the learned dynamics' adaptability and increased robustness to corrupted inputs and changes in the environment, which is highly desirable in challenging real-world robotics scenarios. Video: https://youtu.be/X6k5t7OOnNc.

Citations

• Learning Dynamic Attribute-factored World Models for Efficient Multi-object Reinforcement Learning
• Seeing is not Believing: Robust Reinforcement Learning against Spurious Correlation
• CAR-DESPOT: Causally-Informed Online POMDP Planning for Robots in Confounded Environments
• Bayesian learning of Causal Structure and Mechanisms with GFlowNets and Variational Bayes
• MoCoDA: Model-based Counterfactual Data Augmentation
• Causal Inference for De-biasing Motion Estimation from Robotic Observational Data
• DAGMA: Learning DAGs via M-matrices and a Log-Determinant Acyclicity Characterization
• Causal Imitation Learning with Unobserved Confounders
• Online Dynamics Learning for Predictive Control with an Application to Aerial Robots
• Causal Dynamics Learning for Task-Independent State Abstraction
• Amortized Inference for Causal Structure Learning
• Bayesian Structure Learning with Generative Flow Networks
• CausalAF: Causal Autoregressive Flow for Safety-Critical Driving Scenario Generation
• AdaRL: What, Where, and How to Adapt in Transfer Reinforcement Learning
• The effectiveness of feature attribution methods and its correlation\n with automatic evaluation scores
• Toward Causal Representation Learning
• Do Feature Attribution Methods Correctly Attribute Features?
• MBRL-Lib: A Modular Library for Model-based Reinforcement Learning
• Learning Causal Semantic Representation for Out-of-Distribution Prediction
• Counterfactual Data Augmentation using Locally Factored Dynamics
• Causal Structure Discovery from Distributions Arising from Mixtures of DAGs
• Cascaded Gaussian Processes for Data-efficient Robot Dynamics Learning
• Review of Causal Discovery Methods Based on Graphical Models
• Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models
• Addressing Function Approximation Error in Actor-Critic Methods
• Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
• Proximal Policy Optimization Algorithms
• Axiomatic Attribution for Deep Networks
• An Algorithm for Fast Recovery of Sparse Causal Graphs
• On Curves of Minimal Length with a Constraint on Average Curvature, and with Prescribed Initial and Terminal Positions and Tangents

Related

• Active Causal Structure Learning with Latent Variables: Towards Learning to Detour in Autonomous Robots
• Identifying Causal Structure in Dynamical Systems
• Causal Reasoning in Simulation for Structure and Transfer Learning of Robot Manipulation Policies
• CausalWorld: A Robotic Manipulation Benchmark for Causal Structure and Transfer Learning
• From Continual Learning to Causal Discovery in Robotics