A neural network for target reaching with a robot arm using a stereohead

A self-organizing neural controller for stereohead-robot arm coordination is presented. This neural controller is coupled with a stereohead which implements several neural networks for target representation and control. This control algorithm is based in the DIRECT algorithm which has been developed from a biological inspiration. With this controller a solution to the motor equivalence problem is given. During the initial phase, the model endogenously generates movement commands and activates a correlation process between visual, spatial and motor information that are used to learn its internal coordinate transformations. After learning occurs, the controller is capable of making reaching movements of the arm to prescribed spatial targets using many different combinations of joints. Properties of the controller are compared with psychophysical data on human reaching movements