In this paper, we propose a new motion and force control methodology for constrained robots as an approach of hybrid discrete-continuous dynamical system. The hybrid dynamic system modeling of robotic manipulation tasks with constraints is presented, and the hybrid system control architecture for unconstrained and constrained motion system with parametric uncertainties is synthesized. The optimal reference stiffness of robot manipulator is generated by the hybrid automata as a discrete state system and the control behavior of constrained system which has poor modeling information and time-varying constraint function is improved by the constrained robots as a continuous state system. The performance of the proposed constrained motion control system is successfully evaluated via experimental studies to the constraint tasks.