PUB - Publikationen an der Universität Bielefeld

This dissertation addresses the effects of robot behavior, especially motions, on the user in human-robot applications. When working in such an environment, the operator safety is paramount. Likewise, the user’s perceived safety is essential for well-being, minimizing fatigue effects and ultimately system productivity. Both, occupational and subjective safety must be evaluated for the specific application considering the prevalent working conditions.
As industrial human-robot interaction is task-oriented, its enhancement needs to consider the actual cooperation. This dissertation emphasizes close joint working of humans and robots in a common industrial workspace. A user-centered cooperation is envisaged, including a discussion of human expectations for the robotic assistance. Additionally, relevant characteristics such as use case, interaction specifics, environmental conditions, or system layout are outlined. The interaction with the robot further requires a usable and user-friendly feedback design. It is designed to maintain the user’s attention on the task and avoid distractions in industrial human-robot cooperation. Further, an adaptable motion planner is developed that allows the parametrization of the robot and systematic investigation of the robot motion behavior impact on the worker.
Under controlled robot motion conditions, experiments are performed to investigate what factors need to be considered to make motion planning in an industrial environment as least disruptive as possible. Specifically, the influence of parameters like velocity, acceleration, jerk, distance, approach path, smoothness of motion, and predictability is examined. In a virtual experiment, on the other hand, an opposite approach is followed: The operators themselves can adjust velocity, acceleration, jerk and distance in a way that is comfortable for them. This undertaking also closes a conspicuous knowledge gap: the methodical discussion and consideration of the transferability of results found in a virtual environment to a real application. Furthermore, the question of an automatic determination of the operator workload is investigated. Moreover, it is analyzed whether an increased workload can be assigned directly, i.e. in short-time intervals, to a robot motion event or to an inappropriate parametrization of the robot behavior.
This dissertation shows that even in short time intervals and real scenarios, a workload determination is feasible. It can be used to identify interaction issues in human-robot cooperation. Besides, the applied robot is not capable of exceeding the motion dynamics parameters determined by users to be appropriate; like many other cobots. In addition, especially the parameters distance, oscillations, noise, and timing are identified to be essential. Moreover, the overall consistency of the robot behavior is key for an efficient cooperation.