Toward Integrated Analysis & Testing of Component-Based, Adaptive Robot Systems

Abstract

Recent developments in robotics are heading toward component-based architectures that are able to cope with novel challenges, e.g., increasing autonomy or human-robot-collaboration in Industry 4.0. These applications have in common that the robot system has to adapt to new situations or external disturbances and, thus, has to change or at least to adjust its current task. This development is, among other things, driven by collaborative working environments of robots and humans which pose new challenges for engineering integrated, autonomous robot systems. Especially sensing is an important aspect to safely detect human presence and calculate an according reaction. This includes safe and time-critical reactions to avoid collisions and moreover to evaluate, adapt or even change the current task. Overcoming the separation of working environments and enabling autonomy revises established concepts and techniques that aim at ensuring functional correctness and safety in particular. This (r)evolution has to be taken into account on many different levels, e.g., the revision of safety standards such as ISO 10218 addresses the ongoing development by updating and renewing regulations. The next iteration of these standards will include further restrictions and safety considerations for human-robot-collaboration. In order to comply with these future standards, scalable, automated, and reliable analysis and testing (A&T) techniques will be needed that are able to cope with autonomous, adaptive robot systems. We therefore propose a systematic, model-based approach for analyzing this class of systems: The approach is fully integrated into a robot control system, thus making it possible to address functional quality goals as well as safety aspects within one model-based A&T approach. Robot software in the loop testing is enabled by the use of the S# modeling and analysis framework.