Uncertainty-aware Planning with Inaccurate Models for Robotized Liquid Handling

2025-07-28

AI Summary

To address unreliable planning in deformable object and fluid manipulation caused by inaccurate models—such as simulation-to-reality gaps and poor generalization from limited data—this paper proposes Uncertainty-aware Monte Carlo Tree Search (U-MCTS). U-MCTS integrates physics-based simulation, data-driven dynamics modeling, and explicit uncertainty quantification; it incorporates uncertainty-weighted node evaluation within MCTS to bias search toward high-confidence actions. Evaluated on a liquid pouring task, U-MCTS achieves significantly higher success rates using only a small number of training demonstrations and demonstrates superior robustness over baseline methods. Its core contribution is the first integration of epistemic model uncertainty directly into the MCTS planning loop, enabling more reliable decision-making under distributional shift and substantially mitigating challenges in out-of-distribution generalization and sim-to-real transfer.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Multiagent Systems: Multiagent Systems under Uncertainty
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Physics-based simulations and learning-based models are vital for complex robotics tasks like deformable object manipulation and liquid handling. However, these models often struggle with accuracy due to epistemic uncertainty or the sim-to-real gap. For instance, accurately pouring liquid from one container to another poses challenges, particularly when models are trained on limited demonstrations and may perform poorly in novel situations. This paper proposes an uncertainty-aware Monte Carlo Tree Search (MCTS) algorithm designed to mitigate these inaccuracies. By incorporating estimates of model uncertainty, the proposed MCTS strategy biases the search towards actions with lower predicted uncertainty. This approach enhances the reliability of planning under uncertain conditions. Applied to a liquid pouring task, our method demonstrates improved success rates even with models trained on minimal data, outperforming traditional methods and showcasing its potential for robust decision-making in robotics.

Problem

• Addressing inaccuracies in physics-based and learning-based robotic models
• Improving liquid pouring task reliability under model uncertainty
• Enhancing decision-making in robotics with uncertainty-aware planning

Innovation

• Uncertainty-aware Monte Carlo Tree Search
• Incorporates model uncertainty estimates
• Improves reliability in uncertain conditions