Virtual physically-correct agent’s antagonist muscles neuro-evolutionary control

The virtual agent’s limb link motion control system in the physically-correct virtual environment is considered in this paper. The virtual environment description and the antagonist muscles-driven bone virtual model are presented. The control is brought about with the motoneuron implemented on the basis of the feed-forward neural net-work the weights of which are calculated with the genetic algorithm. The neural network agent’s structure and the learning algorithm together with the motoneuron learning experiments results are given.