Neural computation for controlling the configuration of 2-dimensional truss structure

A method for the manipulator position control was previously proposed by K. Tsutsumi et al. (1987) on the basis of the Hopfield scheme for neural computation. According to the proposed method, the manipulator is modeled not by using polar coordinates but by using the distances between the joint locations alone. The analog neural network realises parallel control by minimizing the total of the energy functions based on the distances, and it can easily treat not only the ordinary manipulator with rigid links but also the elastic arm. The proposed control method with parallelism and flexibility is applied to a two-dimensional truss structure composed of elastic members. Simulation demonstrates that the configuration of the structure can be controlled by adjusting the initial configuration and by tuning the energy balance.<<ETX>>