A framework for anomaly detection of robot behaviors

Autonomous mobile robots are designed to behave appropriately in changing real-world environments without human intervention. In order to satisfy the requirements of autonomy, the robots have to cope with unknown settings and issues of uncertainties in dynamic and complex environments. A first step is to provide a robot with cognitive capabilities and the ability of self-examination to detect behavioral abnormalities. Unfortunately, most existing anomaly recognition systems are neither suitable for the domain of robotic behavior nor well generalizable. In this work a novel spatial-temporal anomaly detection framework for robotic behaviors is introduced which is characterized by its high level of generalization, the semi-unsupervised manner and its high flexibility in application.