Automatic pre-grasps generation for unknown 3D objects

TL;DR

This paper presents an automated method for generating pre-grasps on unknown 3D objects by decomposing objects into parts, encoding free faces, and checking grasp conditions, demonstrated on household objects.

Contribution

The novel approach decomposes objects into primitive parts, encodes free faces, and systematically generates pre-grasps, improving grasp planning for unknown objects.

Findings

Successfully applied on 24 household objects and toys.

Generated grasps considering object part decomposition and face accessibility.

Enhanced grasp planning accuracy with the proposed method.

Abstract

In this paper, the problem of automating the pre-grasps generation for novel 3d objects has been discussed. The objects represented as cloud of 3D points are split into parts and organized in a tree structure, where parts are approximated by simple box primitives. Applying grasping only on the individual object parts may miss a good grasp which involves a combination of parts. The problem has been addressed by traversing the decomposition tree and checking each node of the tree for possible pre-grasps against a set of conditions. Further, a face mask has been introduced to encode the free and blocked faces of the box primitives. Pre-grasps are generated only for the free faces. Finally, the proposed method implemented on a set twenty-four household objects and toys, where a grasp planner based on object slicing method has been used to compute the contact-level grasp plan.