Automatic synthesis of 3D-printable finger design for parallel-jaw gripper

Abstract

With the increasing demand for grasping customised objects, finger design automation is more and more useful in robot industries. Among different kinds of automation approaches, the Generic Optimised Finger Design (GOFD) method is relatively promising. Both single- and multi-function fingers generated by this method can achieve 100% success rate in prescribed pick-and-place tasks. However, this method does not optimise the finger-body design or check the design strength before actual experimental verification. Therefore, this paper studies the Grasp Pose Generator(GPG) function, and proposes a modified pipeline for the GOFD to analyse objects’ geometrical information and generate 3D printable finger designs. Results show that the modified pipeline can shorten the design process time while outputting reliable CAD files. The modified pipeline is implemented in Python, and the FreeCAD and Open3D libraries are used for 3D model generation and analysis.