Chapter 16 Optimal Control of Reaching Movements

Abstract

Summary From computational research, it is elucidated that at least three different problems must be solved to execute visually-guided reaching movements: trajectory planning, coordinate transformation and calculation of motor commands. These problems are ill-posed in the sense that there exists an infinite number of possible solutions. However, the brain easily solves these problems by adopting certain constraints. In this chapter, we discuss optimization principles that define the unique solution for the ill-posed motor control problem; the minimum-jerk model, minimum-torque-change model and minimum-muscle-tension-change model are introduced. Several neural network models are presented to calculate the optimal trajectory and the corresponding motor command. They include Hoff and Arbib's network with an optimal feedback controller, Jordan's sequential network, our cascade network and a new model with forward and inverse models of a motor apparatus.