Recent Cesar research activities in sensor based reasoning for autonomous machines

This paper describes recent research activities at the Center for Engineering Systems Advanced Research (CESAR) in the area of sensor based reasoning, with emphasis being given to their application and implementation on our HERMIES-IIB autonomous mobile vehicle. These activities, including navigation and exploration in a-priori unknown and dynamic environments, goal recognition, vision-guided manipulation and sensor- driven machine learning, are discussed within the framework of a scenario in which an autonomous robot is asked to navigate through an unknown dynamic envi-ronment, explore, find and dock at the panel; read and understand the status of the panel's meters and dials, learn the functioning of a process control panel, and successfully manipulate the control devices of the panel to solve a maintenance emergency problem. A demonstration of the successful implementation of the algorithms on our HERMIES-IIB autonomous robot for resolution of this scenario is presented. Conclusions are drawn concerning the applicability of the methodologies to more general classes of problems and implications for future work on sensor-driven reasoning for autonomous robots are discussed.