Reaching desired states time-optimally from equilibrium and vice versa for visco-elastic joint robots with limited elastic deflection.