Industrial robots revolutionized the car body manufacturing in the automotive industry by reducing heavy manual labor and increasing the quality of car bodies. These improvements are based on precise movement and high payload capabilities of robots, which are much higher than any human can lift and handle properly. In the past decades, the field of robotics expanded towards human-robot collaboration with the idea to combine the strengths of both. Solutions have been developed that are either based on limiting the power and force of the robot to safe amounts around humans or detecting approaching humans early enough to trigger a robot stop such that the robot stands still when a human collides with it. As the first mentioned solution is generally achieved using robots with small payloads, the second one is of interest for big industrial robots. This work focuses on the early human detection solution, which can be applied to robots of all sizes. Many new approaches in research try to improve the safety sensor systems for human robot collaboration of high-payload robots. Unfortunately, the described systems can typically not be applied to real-world applications directly. A major problem is that many improvements are achieved using newly developed safety sensor systems, which need to be certified to be below a given failure rate. The certification process is a major hindrance in developing industrial solutions, especially for research solutions that incorporate new safety sensor technologies. The reason is that the certification requirements of failure safe applications reaches deep into the architecture of sensor data processing, which complicates industrialization of solutions that have not been developed with the functional safety requirements (described in ISO standards) in mind. This work aims to improve human-robot applications by solely relying on existing safety features available in robots and safety sensors today. It therefore has a high applicability in contrast to most solutions that are developed ignoring the state of industrial safety features or relying on the development of new ones. The results show that the introduced methods can improve the layout significantly compared to conventional planning approaches. This is achieved by computing optimized safety layouts, which are utilizing existing safety features in a more sophisticated way than typical planners would use them with their limited work time. This work discusses the whole workflow for computing optimized safety layouts with fenceless robots using a distance monitoring method to ensure safety. It starts with fundamental calculations for generating spaces to limit robot movements and the corresponding spaces for safeguarding using safety sensors. Based on this and considerations regarding the available safety features of robots and sensors today, a general safety layout design is proposed, which can be adapted to multiple use cases. Since it needs to be adapted to bring up its full potential, an optimization problem is formulated, firstly introduced in this work, and ways to solve the problem using population based algorithms. This work concludes with an analysis of the optimization method, determining the effectiveness of the proposed problem representations, and a practical application, which shows how this work can improve the planning process of fenceless industrial robot cells today. Beyond the boundaries of this work, it can be integrated into other problems in the context of automated planning, like Line Balancing Problems or Facility Layout Problems. The improved representation of human-robot related tasks in these problems can yield to a better integration of human-robot applications in factories.
Optimized Safety Layouts for Fenceless Robots
Optimierte Sicherheitslayouts für schutzzaunlose Roboter
2022
Sonstige
Elektronische Ressource
Englisch
BASE | 2022
|Kraftfahrwesen | 1993
|93SF072 Safety on Local Road Layouts
British Library Conference Proceedings | 1993
|Engineering Index Backfile | 1910
|