Semantic State Interpretation For Error Detection

Abstract

Long-term robot autonomy depends on the ability to detect anomalies and flaws since, along with subsequent diagnosis and repair, it enables the robot to reach the necessary levels of robustness and persistency. In fact, if a problem is not quickly identified and fixed, it could cause the robot to damage itself or, in the worst case scenario, cause injury to those nearby. In this thesis, we propose a framework for detecting anomalous behaviors in autonomous robots