Passivity-Based Multilateral Control for Delayed Teleoperation

Kurzfassung

Today, robots are a mature and save tool in automated production lines as well as in collaboration with human workers. Over the past few years, the progress in robotic technologies such as higher computational performance, the development of smarter sensors and impedance controlled light weight robots, has paved the path for robots to new fields of application. Besides space and the nuclear industry, new markets for robots, e.g. in medicine and industrial maintenance have evolved. Since unstructured, dynamic and narrow environments present a challenge for autonomous systems, there is still a high demand for teleoperation systems that give a human operator access to the control of a robot (slave) via haptic input device (master). Besides acoustic and visual feedback, the haptic control loop providing a sense of touch to the operator is crucial in teleoperation systems. Modern control techniques enhanced the transparency, i.e. the quality of immersion into the slaves environment that the human operator perceives via his/her interaction device. Via telemanipulation it became feasible to use robots for plant maintenance or complex constructional tasks e.g. in the nuclear industry. Still, the performance of such so-called bilateral teleoperation systems is limited in several situations. Multiple cooperative robotic agents, autonomous or teleoperated, can achieve a common objective more effectively. Also, in terms of reliability, adaptability and ergonomics multi-robot or multilateral control systems respectively can bring obvious benefits. The basic contribution of this thesis is the development of a modular control framework that allows for an uncomplicated stability analysis for a large variety of multilateral setups thanks to its modularity. The control approach is passivity based which is a widely used stability criterion particularly in the presence of time delay in the communication channel. In a second set of contributions, new control architectures are developed that aim the performance increase in terms of accuracy of position and perceived impedances in the multilateral coupling. In this context, new time domain control approaches for delayed systems, measured force feedback and an extended model-mediated rate control concept are proposed and validated. The third set of contributions of haptic augmentation concepts builds up on these developments. The virtual grasping point concept and a haptic intention augmentation approach are introduced that promise the increase of precision in cooperative tasks and the manipulation of large or flexible objects. Additionally, a role distribution is proposed that promises to increase the system performance in these scenarios and allows for haptic training applications. The work is original in that a large part of the presented approaches brings benefit compared to the state of the art also when applied in standard bilateral setups and in that novel multilateral applications and haptic augmentation concepts are introduced. All discussed concepts are evaluated through real robotic experiments in the course of this thesis and the generality of the modular approach is validated in various applications. Experimental results of task allocation and virtual grasping point concepts, the control of cooperative slave robots as well as rate controlled wheeled mobile robots are presented in real multi-degree of freedom setups. A user study serves the objective evaluation of a set of haptic augmentation approaches. Furthermore, the effects of time delay and a novel haptic intention augmentation approach are demonstrated in a scenario involving a cosmonaut on the International Space Station.