Adaptive Grasping Using an Interphalangeal Flexion Angle Model and Particle Swarm Optimization

In power grasping, all the fingers and thumb are moved simultaneously towards the object centre to form a stable grip. The force imparted on the object while grasping is distributed among all the phalanges. The calculation of interphalangeal flexion angles is essential to ensure their contact with the object surface. For holding cylindrical and spherical shaped objects, the flexion angles follow inverse proportionality with the diameter of the object. In this study, we have proposed a mathematical model by establishing a relationship of the interphalangeal flexion angles with the object diameter to replicate this natural manoeuvre in a hand prosthesis. We have derived that the sum of tangents of all the 14 interphalangeal flexion angles involved in power grasps, depends only on the length of intermediate phalanx of all the fingers and the object diameter. This relation eliminated the requirement of other phalangeal lengths and thus reduced overall variable complexity. To automate the computation of interphalangeal flexion angles, here we have implemented particle swarm optimisation (PSO). The relationship of the joint angle variation with the object diameter is used here as the fitness function. The resulted flexion angles were further evaluated for their efficacy in a simulated hand grasping model. In contrast to the generic prosthetic hands, where the joints are sequentially rotated according to their constraints from the object surface, this model allows simultaneous rotation of the joint angles according to the optimum fitness function using PSO.