Abstract: A method based on quadratic performance index is proposed to plan self-motion trajectories of planar redundant manipulators.As physical joint limits exist in all actual redundant manipulators,avoidance for joint limits and joint velocity limits is taken into account in the presented self-motion-planning scheme.A primal-dual neural network(PDNN)based on linear variational inequalities(LVIs)is introduced as the real-time solver for the resultant quadratic programming scheme. Simulation results verify the validity of the neural-network-based self-motion-planning scheme.