This thesis focuses on the development of new motion and force control methods for flexible joint robots based on joint torque feedback.

Motion control issues of flexible joint robots are addressed first. A two-stage control scheme, consisting of a motion controller and a joint torque controller, is established in a systematic way for controlling flexible joint robots in the general n-link case. To deal with uncertainties in the robotic system, an adaptive and robust control algorithm is developed assuming that all system parameters, including the joint flexibility values, are unknown except for some of their bounds. The system stability is analyzed via the Lyapunov stability theory. Compared with published work, the result has the distinct feature that it does not require restrictions on joint flexibility, nor an exact knowledge of the parameters of the joint actuator subsystem.

Simultaneous motion and force control of flexible joint robots in constrained motion is then investigated. Three adaptive and robust control methods are developed to control both the motion and contact force simultaneously. The first two methods are based on a reduced constrained dynamic model;​ the first is an adaptive and robust saturation control method that accommodates parametric uncertainty in the system dynamics and guarantees the uniform ultimate boundedness of tracking errors;​ the second is an adaptive and sliding control algorithm that can cope with both parameter uncertainty and unknown additive bounded disturbance, while achieving global asymptotic stability. The third method, based on the derivation of a generalized transformed dynamic model, is a unified control scheme which controls flexible joint robots in terms of general coordinates, and ensures the global asymptotic convergence of motion, force and joint torque tracking errors, with out requiring persistent excitation. Distinct from previous work reported in the literature, the proposed control schemes using a two-stage control strategy provide systematic approaches for motion and force control of flexible joint robots in the general n-link case, without requiring an exact knowledge of robot dynamics.

Coordinated control of multiple flexible joint robots is also addressed in this work. The proposed two-stage control scheme is extended to motion and force control of multiple flexible joint robots holding a common object. In the control scheme, new adaptive coordinated controllers are developed for both free and constrained motion. The controllers control the motion, internal force and contact force simultaneously, without requiring an exact knowledge of the dynamic parameters of the multirobot system. Asymptotic convergence of the control system is proved theoretically. Compared with previous work in the literature, the proposed control scheme provides a new approach for dealing with the uncertainty in the multiple flexible joint robot system.

Simulations have been conducted extensively to test the effectiveness of the proposed control methods. In addition, experimental implementation of the proposed adaptive and robust motion control scheme has been carried out on the IRIS robot manipulator. Based on the comparison with different control methods, the experimental results clearly illustrate the effects of joint flexibility on the control systems and verify that the proposed control scheme can control the flexible joint robot stably, in spite of dynamic uncertainty in the system. The theoretical developments are confirmed by simulations and experiments.