REAL-TIME DIAGNOSTICS FOR ROS RUNNING SYSTEMS BASED ON PROBABILISTIC PATTERNS IDENTIFICATION

Abstract

Autonomous mobile robots consists of various software modules to achieve given goal, including solving complex navigation tasks as localization, mapping or path planning. These tasks are highly dependent on the quality of data measured and gathered from hardware subsystems. Using Robot Operating System (ROS) as integration basis reduces the development effort and time to market. While ROS framework itself is considered as reliable and stable to run even soft real-time tasks, in case of any internal failures on data misreadings can be problematic to debug or even identify the problem for common user. Due to this unpleasant situations we develop a virtual assistant, internally represented as diagnostic expert system, to help users to identify and possibly fix the problem.