Risk Aware Robots - Health Estimation and Capability Selection

Abstract

By not knowing their limits, field and service robots either act overly careful, limiting their potential, or are oblivious to risks involved in their mission, leading to possibly dangerous behaviors. We present our Health-Tree approach to enable robots and their operators to intuitively estimate the robots current well being. In combination with our Skill-Tree the consequences of using robot capabilities are made visible, enabling risk aware decisions. The system was tested with two different robot types, LAURON V and a bebop UAV.