Verification and validation of robotic systems's safety

Robotic systems are widely used in many fields, especially in industrial production, thanks to their increasing capabilities with sensor and software technologies. While they completely replace people in some tasks, they can work with people in some tasks. Therefore, an error that may occur in very complex systems such as robotic systems can affect not only itself but also its surroundings. This raises serious safety concerns. Thus, it is critical to utilize verification and validation in the software development of robotic systems to ensure that they perform expected behaviors. In this study, a Model-Aided Runtime Verification Tool (MARVer) tool is proposed, which implements a model-based specification and verification method for robotic systems. The objective of the proposed approach is to enhance the reliability, reusability, and ease of verification for robotic systems by merging formal verification techniques, specifically model checking and runtime verification. The combination of these techniques leads to a more stable and rigorous process for evaluating the safety of robotic systems. Therefore, the proposed method provides a comprehensive solution to address the challenges associated with verifying complex robotic systems. Furthermore, MARVer offers an interface for managing the verification process, eliminating the need for complex terminal operations. This feature streamlines the verification process and allows users to easily oversee and manage the evaluation of robotic systems. Within the scope of the study, experiments are conducted to show the capabilities of MARVer. A specific test case was conducted to evaluate the system's ability to avoid obstacles safely while following pre-defined trajectory plans. In the experiment, a robot that took pictures at certain points by walking around a bus body-in-white was verified for safety. It was verified whether the determined safety distance threshold value was exceeded by observing the distance data taken from six different points during the movement of the robot. In the experiment, both simulation and real environment tests were carried out. The study's results indicated that the proposed method and tool are effective in verifying the safety of robotic systems.