With recent advances in processors, developed control algorithms can eliminate some instruments and consequently simplify the overall system and reduce the cost. One of the expensive sensors that can be eliminated from the processes is force sensor. According to developments in robot applications and their participating in human society, position control without considering interactions with environment is not desirable. In many tasks, robot is in touch with different surfaces and moves on them. It produces an unknown variable force interacting with end effector. Furthermore position controlled robots may damage themselves, environments or humans to reach their desired positions. Therefore, measuring this interactive force is necessary for control, analysis or decision purposes. Mounting sensors on the end-effector for sensing this force was the only way for years, which led to high cost of repairs and maintenance. High noise, soft structure, limited dimensions of sensing and increasing complexity are the other drawbacks of force sensors. In this thesis, several approaches are presented to estimate the external forces which are exerted on the manipulator. Then they are developed to be used as an implicit compliance control where the adaptive estimators are used to estimate and compensate disturbing forces in adaptive position tracking control and an outer force control loop is used to modify the trajectory of the manipulator. These approches are adaptive force estimator, parameter estimator of linear environment and function estimator of nonlinear environments using wavelet neural network respectively. All of them are designed to estimate the external forces and the adaptive laws update their parameters. In addition, these estimations are used in several force control methods, such as hybrid force/position control and impedance control as well. A Lyapunov based design offers a force control law that modifies the desired trajectory of the manipulator and an adaptive position/velocity control law that tracks the modified trajectory and estimates force and robot parameters. The ability of tarcking force control of manipulators are possible for environment estimators, which estimate the stiffness of environment in addition to estimating the force. An auxiliary term is added to position/velocity control loop that improves the performance of force control in contact with the environment. This auxiliary term is useful for all implicit force control algorithms, either for approaches with sensor or for sensorless approaches. The proof of stability of force control for the interacted manipulator is presented. Moreover, for robots with non-contact operation it is guaranteed that the robot reaches to the environment. Unlike the other sensorless methods in force control that use disturbance observer and need an accurate model of the manipulator, in this method, the unknown parameters of the robot can be estimated along with the force control. The other important limitation in disturbance observers is the necessity of computing inverse of Jacobian traose to reach the force values, which cause problems for the system in the vicinity of robot singular points. In the proposed method the jacobian matrix is in adaptive laws and there is no need to compute its inverse. For validating the results some simulations of a 2-DOF planar manipulator are performed. They show the efficiency of the proposed controller in dealing with different environments and in presence of uncertainty. Keywords: Adaptive control, Robot manipulator control, Sensorless force control, Lyapunov based design, Adaptive force and parameters estimator, Hybrid force/position control, Impedance control. ?
