Sensor-based navigation of air duct inspection mobile robots

12 This paper deals with an image sensor system and its position estimation algorithm for autonomous duct cleaning and inspection mobile robots. For the real application, a hierarchical control structure that consists of robot motion controller and image sensor system is designed considering the efficient and autonomous motion behaviors in narrow space such as air ducts. The sensor's system consists of a CCD camera and two laser sources to generate slit beams. The image of the structured lights is used for calculating the geometric parameters of the air ducts which are usually designed with a rectangular section. With the acquired 3D information about the environment, the mobile robot with two differential driving wheels is able to autonomously navigates along the duct path without any human intervention. For real time navigation, the relative position estimation of the robot are performed from 3D image reconstructed by the sensor system. The calibration and image processing methods used for the sensor system are presented with the experimental data. The experimental results show the possibility of the sensor based navigation which is important for effective duct cleaning by small mobile robots.