Flexible Runtime Monitoring Support for ROS-based Applications

ROS-based robotic applications are becoming increasingly common in various different application domains, performing diverse tasks. Examples include autonomous vehicles, small unmanned systems, as well as applications of Cyber-Physical Systems. What all these systems have in common is their tight integration between hardware and software components, and close interactions with humans, e.g., on a shop floor, or autonomously driving robots as part of a warehouse system. This, in turn, requires monitoring the behavior of the system at runtime and ensuring that it behaves according to its specified requirements. However, establishing and maintaining runtime monitoring support is a non-trivial task, requiring significant up-front investment and domain knowledge. To alleviate this issue, this thesis identifies the monitoring challenges of ROS-based applications and their corresponding requirements for a monitoring framework. Based on these requirements, ROMoSu is introduced, a flexible runtime monitoring framework for ROS-based systems that allows defining multiple scenarios, or application-specific configurations, taking into account different monitoring needs (e.g., different properties and monitoring frequencies) and provides editors for creating, maintaining, and managing configurations at runtime. As part of an evaluation, experiments with three different use cases are conducted, including physical and simulated applications. Results confirm that ROMoSu can be successfully used to create monitoring configurations with little effort, create efficient monitors, and perform constraint checks based on the collected runtime data.