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

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 3× 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.

We propose CURA-PPO (Collision Uncertainty-Risk Aware PPO), a reinforcement learning framework designed to ensure safe non-prehensile manipulation under object-induced occlusion. Our approach utilizes a Distributional Collision Estimator (DCE) that takes partial observations—augmented by confidence maps to track sensor reliability—and predicts collision possibility as a distribution rather than a single value. By extracting the risk (mean) and uncertainty (variance) from this distribution and incorporating them as intrinsic costs into the policy objective, the robot learns to balance task progress with active perception. This enables the mobile manipulator to proactively adjust its viewpoint to resolve occlusions and reveal hidden obstacles while pushing the object.