Accuracy Improvement and Joint Stiction Relieve in Robot Arms Movement Applying Torque-Based Cartesian Impedance Control With Dithering

Abstract: Robots are nowadays present and essential in a wide range of applications, and robot arms in particular are capable of performing various tasks such as inspection, material handling, and welding. They can be classified into several types, depending on their structure and purpose. This project focuses on two collaborative robot arms, a type that is designed to work in the presence of human interaction. They both possess a manipulator at the end of the arm known as end effector, which normally realizes the tasks and is responsible for following trajectories instructed by the controller. One of the challenges faced by robot arms is static friction in the joints, also known as stiction. This phenomenon generally occurs in low-speed regimes, having its main component expected at zero velocity, and might affect the robot’s motion and therefore its accuracy. During this thesis work, stiction will be characterized through different approaches and its impact will be mitigated by applying a technique known as dithering. Dithering consists of adding a periodic signal, in this case a sinusoidal wave, to the robot controller in order to reduce the stiction effect. A particular dithering signal can be built for each of the robot joints since their dynamics are also different between them. Two approaches were designed to characterize the stiction. The first approach involves the identification of the robot dynamics, while the second approach, referred to as the Single Joint Experiment (SJE), is an empirical iterative method in which the robot arm conducts individualized motions for each joint. Subsequently, dithering signals were built after characterizing the stiction bands, and applied to different motion experiments. The SJE was first improved by reducing the minimum torque needed to surpass stiction, and at the same time this procedure was used to obtain the optimal parameters for the joints’ dithering signals. Finally, applying these optimal dithering signals to the controller, the deviations in the trajectories were also reduced, resulting in improved accuracy.