Adaptive Position Feedback Control of Parallel Robots in the Presence of Kinematics and Dynamics Uncertainties

Uncertainties in the kinematic and dynamic parameters of a parallel robot are unavoidable. The problem is more crucial in the cases where the manipulator interacts with the environment and when it is large–scale or deployable. Furthermore, precise measurement of the velocity of the end-effector is almost inaccessible in practice. This paper addresses the above shortcomings by designing of an adaptive trajectory tracking controller with merely position feedback of joint and task space variables. Simplicity of implementation, separation of adaptation laws of dynamic and kinematic parameters, and reduction of the number of adaptation laws such that in some cases, e.g., cable-driven robots, it is identically equivalent to the number of unknown parameters are some advantages of the proposed controller. The method’s efficiency is shown via implementation on a cable-driven parallel manipulator and an intraocular surgery robot. Note to Practitioners—In order to achieve a suitable response in parallel robots with traditional controllers, a precise knowledge of kinematic and dynamic parameters, together with accurate measurement of velocity, is required. In practice, these values are usually derived by robot calibration and identification. Since the system’s parameters may alter or depend on external factors such as temperature, these time-consuming methods should be implemented repeatedly while the robot is out of duty. Additionally, precise velocity measurement is a prohibitive task and requires expensive instruments. This article presents a controller to address the above shortcomings. By this means, a simple adaptive controller based on position feedback is designed such that via suitable estimation of kinematic and dynamic parameters, trajectory tracking is obtained without the need for accurate initial estimates of the parameters. The proposed method has a number of advantages, including separation of the adaptation laws of kinematic parameters and dynamic parameters, simple representation of the Jacobian matrix in regressor form, and there is no requirement for velocity feedback. Furthermore, a force distribution method is introduced for redundant robots. Hence, the proposed controller is an appropriate alternative to traditional controllers widely used in industries.