Center-of-Mass-based Robust Grasp Pose Adaptation Using RGBD Camera and Force/Torque Sensing

TL;DR

This paper introduces a method for improving robotic grasp stability by estimating object center-of-mass using joint torque sensors and RGBD data, without extra sensors or extensive learning.

Contribution

It presents a novel approach to adapt grasp poses based on center-of-mass estimation, enhancing robustness without additional tactile sensors or large-scale training.

Findings

Effective in reducing object dropping in simulations

Improves grasp stability with simple sensor data

Validated in Mujoco simulation environment

Abstract

Object dropping may occur when the robotic arm grasps objects with uneven mass distribution due to additional moments generated by objects' gravity. To solve this problem, we present a novel work that does not require extra wrist and tactile sensors and large amounts of experiments for learning. First, we obtain the center-of-mass position of the rod object using the widely fixed joint torque sensors on the robot arm and RGBD camera. Further, we give the strategy of grasping to improve grasp stability. Simulation experiments are performed in "Mujoco". Results demonstrate that our work is effective in enhancing grasping robustness.