Autonomous industrial tank floor inspection robot

Abstract

Many storage tanks used in industry are inspected manually in order to identify leakage and corrosion. This repetitive task can harm the operator as the operator has to expose themselves to danger inside the tank. The environment of the tank can be harmful because it might contain highly hazardous substances. Additionally, the manual inspection process also requires highly skilled operators for precision. This project proposes the use of an autonomous inspection robot to overcome this problem. The proposed robot is based on the TurtleBot with the Microsoft Kinect camera as a distance sensor. The MinnowBoard MAX board is used as the computing platform running the Robot Operating System (ROS) for tank floor mapping and navigation of the robot. The GMapping Simultaneous Localization and Mapping (SLAM) algorithm is used to do perform the mapping process. Evaluation was performed by comparison against physical measurement of the foyer of the Faculty of Engineering and Built Environment (FKAB) at Universiti Kebangsaan Malaysia (UKM). The results show that the GMapping algorithm is able to generate relatively accurate maps even with the low cost platform and will perform even better with more accurate sensors.