Autonomous stair identification, climbing, and descending for tracked robots

This paper presented a controller for autonomous stair identification, climbing, descending for a tracked robot. The controller relies on stair features extracted from the perception system and a smooth feedback steering control based on fuzzy logic. Experimental results indicate that the proposed method offers robust performance without the need for an accurate attitude estimator nor a detailed model of the robot´s physical and dynamics properties. Experimental results have verified that the controller is sufficient for stair climbing and descending task in urban environments. Hence, the potential to be easily implemented on any tracked vehicles that are capable to be manually driven in a stair.