Linking Sensed Images to an Ontology of Obstacles to Aid in Autonomous Driving