Pipeline Robot Positioning System Based on Machine Learning

Abstract

With the continuous development of artificial intelligence, sewage pipeline robot is also gradually intelligent. This intelligent system is inseparable from machine perception systems and machine learning. Therefore, for the problem that the robot in the sewage pipeline can not locate accurately, a pipeline robot positioning system based on machine learning is designed. From the perspective of computer vision, the full convolution neural network is used to locate the robot. The robot can realize its positioning function by acquiring a single RGB (Red Green Blue) image from the current perspective. The positioning results are combined with the robot mobile platform system to complete the robot navigation task. Through the test in the simulated sewage pipeline scene, the practical value of the system method is verified. The experimental data show that the positioning and navigation system has high positioning accuracy, strong stability and certain practical value.