Inverse kinematic solution for a 7 DOF robot with minimal computational complexity and singularity avoidance

One of the problems in real-time control of redundant manipulators is considerably increased computational complexity compared with nonredundant robots. An attempt is made to reduce computational complexity by combining analytical and pseudoinverse solutions. In this way the dimensions of the Jacobian are considerably reduced, yielding several times less computational complexity. Further reduction of computational time is achieved by applying the gradient projection method to the actually redundant subrobot. A simple performance criterion for avoiding wrist singularity is proposed.