A MODEL FOR LEARNING COORDINATED FAST MOVEMENTS

  • M. Gruber
  • A. Kraft
  • A. Golhofer
Keywords: neural network, motor control, coordination, feed forward controller

Abstract

Coordinated fast movements are characterised by an effective intermuscular interaction. The result of this intermuscular coordination is a straight path towards a given target. The speed profile of the movement is "bell-shaped", so that a movement with a smooth stop results (an overshoot is not observed). In this work an artificial neural network acts as a controller of an idealised human arm during a catching movement. The model arm is taken from the literature with minor changes. The nervous system is modelled by an artificial neural network (ANN). It consists of a sensory map that is connected to a motor map by an intermediate associative layer. The results demonstrate that simple neural networks in interaction with musculoskeletal dynamics are able to model the ability of the central nervous system to coordinate fast movements.