Autonomous Sewer Robot: A Laser Marker-Based Detection System

Abstract

Navigation technologies are becoming more advanced, helping to solve complicated problems in various fields. Navigation can be classified as global, in which predefined data and reference points from the working environment are used to generate a path, and local, where the map is generated momentarily by acquiring data from outside using sensors. As navigation tasks become more demanding, working environments can become very complicated with an increasing number of dynamic obstacles or, in some cases, a lack of global references, which may have particularly notable impacts on communication. Inspection robots that are required to work in underground sewers are often required to work completely locally, relying on sensor data. For this reason, in this study, a cost-efficient laser marker-based obstacle detection and measurement system is designed and tested for future use in autonomous sewer robot local path generation. Our experiments show the convenience of applying four linear laser markers with an RGB camera to fully inspect upcoming obstacles in the region covering the front robot dimensions. The results show distance measurement accuracy of up to ±1.33 mm and obstacle width accuracy of up to ±0.28 mm measured from a 130 mm range. Nevertheless, accuracy is strongly dependent on the segmentation of image resolution, lighting, and reflection of the inspected surfaces. Also, it depends on the configuration of laser markers and the RGB camera position.