ABSTRACT
This chapter focussed on developing an optimal redundancy resolution scheme for visually controlled redundant manipulator. The redundancy is resolved optimally while reaching the positions defined in the vision space. The SNAC based redundancy resolution scheme proposed in the previous chapter has been extended to the vision space with similar dynamical system formulation. Two novel neural network architectures have been proposed to cope with the challenges occurring while defining the control trajectories in the vision space. The proposed neural network architectures are tested both in simulation and real-time on the PowerCube manipulator for trajectory spanning the entire vision space. Finally the critic based redundancy resolution scheme is successfully used to control the PowerCube manipulator mounted with Barrett Hand, to grasp the ball located in various positions within the workspace.
