Autonomous Inspection of Complex Environments by Means of Semantic Techniques

The autonomous inspection of complex environments is a challenging task. An autonomous inspection robot should actively examine entities of interest (EOIs), e.g. defects, and should perform additional inspection actions until the data analysis results reach an appropriate level of confidence. In this paper a semantic approach for inspection planning, plan execution, assessment of the data analysis results, decision making and replanning is proposed. The main idea is to incorporate human expert knowledge via a semantic inspection model. For the experimental evaluation of this approach the detection and classification of waste on irregular terrains with the hexapod walking machine LAURON is chosen. First preliminary simulation results are presented.