HURC-MM: Holistic Uncertainty-Responsive Control for Mobile Manipulator Grasping
Shangkun Dai, Qing Tang, Maoxin Yang, Xiang Yang, Huafeng Chen, Qun Wang
Abstract
We present a holistic uncertainty-responsive control system that enables mobile manipulators to perform human- like adaptive grasping by dynamically coordinating base-manipulator motion and uncertainty-aware decision-making. Our key innovations include: (1) a continuous visual monitoring strategy: Distant Watch Close Grasp (DWCG) for dynamic object grasping that maintains persistent visual tracking throughout the entire approach-to-grasp trajectory which ensures uninterrupted target observation from initial detection to final grasp execution; (2) a strategic active retreat mechanism that optimizes withdrawal trajectories based on uncertainty quantification to address Blurred-View Clear-Grasp (BVCG) challenges; and (3) a unified control framework that integrates DWCG and BVCG strategies via quadratic programming. By weighting distance-to-target and uncertainty metrics, our method dynamically optimizes end-effector pose - DWCG guides orientation while BVCG determines position. This enables seamless approach/retreat transitions within a single control loop, eliminating decoupled planning. Experimental validation demonstrates that HURC-MM significantly improves grasping reliability and adaptability in dynamic environments through its human- like “observe-while-moving” capability and intelligent retreat from tracking uncertainty.
BibTeX
@inproceedings{ral2026_hurcmmholisticun,
title = {HURC-MM: Holistic Uncertainty-Responsive Control for Mobile Manipulator Grasping},
author = {Shangkun Dai and Qing Tang and Maoxin Yang and Xiang Yang and Huafeng Chen and Qun Wang},
booktitle = {RA-L 2026},
year = {2026}
}