Optimal motion generation of a flexible macro-micro manipulator system using genetic algorithm and neural network

In this paper, a new approach to solve the inverse kinematics of a flexible macro-micro manipulator system is proposed. The macro-micro manipulator system consists of a macro flexible manipulator, and a micro rigid manipulator which is used to compensate for the errors at the tip of the system. Apparently, such a macro-micro system is a redundant system, of which the inverse kinematics remains challenging, with no generic solution to date. Here, optimal joint motions, namely the manipulator system configuration, are generated using a genetic algorithm and a neural network. A fitness function is dedicated to the genetic algorithm to optimize the discrete solution of the inverse kinematics of the manipulator system. Then the discrete solution is further generalized by a forward neural network. A new compensability measure is defined in this paper. The proposed approach shows excellent performance on error compensation, as demonstrated by the simulation results.