Neurohand Solving the Inverse Cinematic of an Anthropomorphic Arm

When an anthropomorphic arm has to reach a point in its workspace, many joint configurations are possible. That is the problem of inverse cinematic redundancy. This problem consists on several possible arm joint configurations for reaching the target point with the wrist (open cinematic chain). The humans solve the cinematic redundancy in a natural way learned in childhood. In this paper we describe the learning algorithm for artificial neural networks used to solve the cinematic redundancy in order to make a virtual robotic anthropomorphic arm has a ‘human’ joint configuration to reach a target point.