In this paper, we present a no vel method of mapping the task space on
to the joint space. This map divides the joint space into patches (dis
jointed regions) and an n ways accessible region in the task space is
exactly covered by n patches. This mapping allows one to predict the t
ask space paths corresponding to the joint space paths without solving
inverse kinematics along the path. The map is then utilized for joint
space planning. Numerical examples are presented for illustration.