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

TL;DR

The paper tackles object-induced occlusion during non-prehensile manipulation with onboard sensing by introducing CURA-PPO, which models collision risk as a distribution and extracts both expected risk and predictive uncertainty to guide actions. A Distributional Collision Estimator (DCE) predicts the distribution of future collisions using quantile regression, providing risk and uncertainty signals that are integrated into a CURA-PPO objective with two intrinsic costs. The approach promotes active perception while maintaining task progress, demonstrated through extensive simulations showing up to a 3x improvement over baselines across varying object sizes and obstacle configurations. The work advances safe autonomous manipulation in cluttered environments using only local sensing and highlights future real-world deployment and richer sensing modalities.

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.

Figures (7)

• Figure 1: Uncertainty-Aware Non-prehensile Manipulation under Object-Induced Occlusion. A mobile manipulator pushes an object to the target using local onboard sensing. (1) The object induces occlusion, hiding an unforeseen obstacle. (2) The mobile manipulator adjusts its behavior to actively reduce uncertainty from occlusion, revealing the hidden obstacle for safe non-prehensile manipulation.
• Figure 2: Overall Framework. Local observations including proprioceptive, object, goal, and confidence map features are processed by the policy network $\pi_{\phi}$ (see Sec. \ref{['sec:push']}), which outputs velocity commands executed through differential inverse kinematics. The Distributional Collision Estimator (DCE, $f_{\varphi}$) predicts collision possibility as a distribution, extracting risk and uncertainty signals (see Sec. \ref{['sec:dce']}) that augment the PPO objective in CURA-PPO, encouraging behaviors that reduce both collision risk and perceptual uncertainty (see Sec. \ref{['sec:cura']}).
• Figure 3: Environment used for training and evaluation. The initial pose for the object is sampled from the red region. Obstacles are randomly generated within the yellow region, and the goal pose is sampled from the green region. The object and obstacles are cuboids of varying sizes, with 0 to 6 obstacles present in each episode. The top-left shows an obstacle, and the top-right shows the object.
• Figure 4: Qualitative Manipulation Behavior under Occlusion. Trajectories shown with robot paths in green and object paths in magenta. (Left) CURA-PPO with corresponding confidence maps. The robot actively repositions laterally (a→b, c→d) to reveal occluded regions, successfully detecting and avoiding obstacles during pushing the object to the goal. (Right) The baseline policy, which lacks both confidence maps and DCE-based costs, fails to explore occluded regions, resulting in collision with an undetected obstacle that was hidden behind the manipulated object (e→f). Without these uncertainty-aware mechanisms, the robot maintains a direct path that leads to task failure.
• Figure 5: Visualization of DCE predictions at time steps from Fig. \ref{['fig:qualitative']}. (Left) CURA-PPO reduces variance from c[red] to d[blue] through active perception. (Right) Baseline maintains high variance from e[red] to f[blue], leading to collision.