Analysis of Inverse Kinamtics of an Anthropomorphic Robotic hand

In this paper, a new method for solving the inverse kinematics of the fingers of an anthropomorphic hand is proposed. Solution of inverse kinematic equations is a complex problem, the complexity comes from the nonlinearity of joint space and Cartesian space mapping and having multiple solutions.This is a typical problem in robotics that needs to be solved to control the fingers of an anthropomorphic robotic hand to perform tasks it is designated to do. With more complex structures operating in a 3-dimensional space deducing a mathematical soluation for the inverse kinematics may prove challenging. In this paper, using the ability of ANFIS (Adaptive Neuro-Fuzzy Inference System) to learn from training data, it is possible to create ANFIS network, an implementation of a representative fuzzy inference system using ANFIS structure, with limited mathematical representation of the system. The main advantages of this method with respect to the other methods are implementation is easy, very fast and shorter computation time and better response with acceptable error.