Uncertainty-Aware Non-Prehensile Manipulation with Mobile Manipulators under Object-Induced Occlusion

TL;DR

This paper introduces CURA-PPO, a reinforcement learning framework that enables mobile manipulators to perform safe, effective non-prehensile manipulation under object-induced occlusion by modeling uncertainty and actively gathering information.

Contribution

The paper presents a novel uncertainty-aware RL approach that explicitly models collision risk under partial observability, improving manipulation success in occluded environments.

Findings

CURA-PPO achieves up to 3X higher success rates than baselines.

The method enables active perception to resolve occlusions.

Robust manipulation performance across various object sizes and obstacle configurations.

Abstract

Non-prehensile manipulation using onboard sensing presents a fundamental challenge: the manipulated object occludes the sensor's field of view, creating occluded regions that can lead to collisions. We propose CURA-PPO, a reinforcement learning framework that addresses this challenge by explicitly modeling uncertainty under partial observability. By predicting collision possibility as a distribution, we extract both risk and uncertainty to guide the robot's actions. The uncertainty term encourages active perception, enabling simultaneous manipulation and information gathering to resolve occlusions. When combined with confidence maps that capture observation reliability, our approach enables safe navigation despite severe sensor occlusion. Extensive experiments across varying object sizes and obstacle configurations demonstrate that CURA-PPO achieves up to 3X higher success rates than the baselines, with learned behaviors that handle occlusions. Our method provides a practical solution for autonomous manipulation in cluttered environments using only onboard sensing.