In this paper, a variable structure control scheme for constrained robots is developed. This control law ensures the asymptotic convergence of the position errors to zero and also the boundedness of the force tracking error. The control scheme requires a few online computations and thus can be easily implemented. The proposed control law does not require measurements of the joint accelerations or the estimation of the derivative of the force. Experimental results are presented to verify the theoretical developments.Assumption 1: The robot is not in singular positions; therefore the Jacobian matrix is invertible.
Assumption 2:The constraint surface is frictionless, and the kinematic model of the constraint function F…q † is known and twice continuously diVerentiable.
Assumption 3:The contact between the robot and the constraint surface is rigid.
Assumption 4:The desired trajectories q d , _ q q d , q q d and ¶ d are bounded for all t.Remark 1: The desired trajectory q d is designed using the kinematic model of the system while keeping away