Minimum infinity-norm kinematic solution for redundant robots using neural networks

The aim of the paper is to develop a new method of applying computational intelligence for determining a minimum infinity-norm solution to the velocity inverse kinematics problem of redundant robots i.e. computing a joint velocity vector whose maximum absolute value component is minimum among all possible joint velocity vectors corresponding to the desired end-effector velocity. A fully neural-network-based (Tank-Hopfield network) computational scheme is proposed for its implementation. At each time step, the neural network produces both the least-norm joint velocity solution and the infinity-norm solution. Simulation results demonstrate that the proposed method is effective.