Recursive recurrent neural network: A novel model for manipulator control with different levels of physical constraints

Manipulators actuate joints to let end effectors perform precise path tracking tasks. Recurrent neural network which is described by dynamic models with parallel processing capability, a powerful tool for kinematic control of manipulators. Due physical limitations and actuation saturation manipulator joints, the involvement joint constraints manipulators essential critical. However, current existing methods based on recurrent networks mainly handle limited levels angular constraints, best our knowledge, higher order are not yet reported. In this study, first time, novel recursive model proposed solve issue different can be formulated as new manifold system ensure solution within all in orders. The theoretical analysis shows stability purposed its convergence solution. Simulation results further demonstrate effectiveness method end-effector under Kuka system. Comparisons other such pseudoinverse-based conventional substantiate superiority method.