Learning to Efficiently Plan Robust Frictional Multi-Object Grasps

TL;DR

This paper introduces a neural network-based method for planning robust multi-object grasps with friction, significantly improving efficiency and success rates in robotic decluttering tasks involving multiple objects.

Contribution

It presents a novel approach incorporating friction into multi-object grasp planning and trains a neural network to enhance robustness and efficiency in decluttering scenarios.

Findings

13.7% increase in success rate

1.6x increase in picks per hour

6.3x decrease in grasp planning time

Abstract

We consider a decluttering problem where multiple rigid convex polygonal objects rest in randomly placed positions and orientations on a planar surface and must be efficiently transported to a packing box using both single and multi-object grasps. Prior work considered frictionless multi-object grasping. In this paper, we introduce friction to increase the number of potential grasps for a given group of objects, and thus increase picks per hour. We train a neural network using real examples to plan robust multi-object grasps. In physical experiments, we find a 13.7% increase in success rate, a 1.6x increase in picks per hour, and a 6.3x decrease in grasp planning time compared to prior work on multi-object grasping. Compared to single-object grasping, we find a 3.1x increase in picks per hour.