An adaptive control approach is presented to control the position and force of flexible joint robot manipulators interacting with rigid environment when the robot parameters are uncertain and an observer is used to estimate the state vector used in the feedback.

A feedback linearizable fourth order model of the flexible joint robot is constructed. Also, a constraint frame is derived which enables decoupling of the system into a position subsystem and a force subsystem.

A model-based and an adaptive control approaches are presented. In the model-based algorithm, the robot parameters are assumed to be known and full state feedback is available using the robot dynamic model. It is shown that a nonlinear feedback control law can be designed which linearizes the system, uses the constraint frame to decouple the system into the position and the force subsystems, and imposes desired closed loop characteristics in each subsystem independently. In the adaptive control algorithm, the robot parameter are uncertain and only link position is available. This control algorithm consists of a nonlinear control law which has the same structure as the one derived for the model-based algorithm except that it uses the estimated parameters and observed state, an adaptation law and, a sliding mode observer. The sufficient conditions under which the closed loop system composed of the flexible joint robot interacting with a rigid environment, the observer, the controller, is stable are presented.

An experimental two link flexible joints robot manipulator constrained by a straight rigid wall is used to evaluate the two control algorithms.