Development of a Robotic Hand Using Bioinspired Optimization for Mechanical and Control Design: UnB-Hand

Abstract

For the last four decades, the development of robotic hands has been the focus of several works. However, a small part of those approaches consider the exploitation of parallelism of FPGA-based (Field Programmable Gate Arrays) systems or discuss how using bioinspired optimization algorithms could improve the mechanical and controller components. This work considers developing a bioinspired robotic hand that achieves motion and force control with a logic hardware architecture implemented in FPGA intended to be replicated and executed with suitable parallelism, fitting a single device. The developed robotic hand prototype has five fingers and seven DoF (Degrees of Freedom). Using bioinspired optimization, such as PSO (Particle Swarm Optimization), both the rigid finger mechanism and the impedance controller were optimized and incorporated the results in several practical grasping experiments. The validation of this work is done with the Cutkosky grasping taxonomy and some grasping experiments with interference. The tests proved the proficiency of this works for a wide range of power and some precision grasp. The reader can see the experiments in the attached videos.