Environment Prediction to Enhance the Navigation System of Water Pipeline Inspection Platforms

The water distribution network is one of the critical buried infrastructures of a nation. Failure of any water pipes or water mains can disrupt the everyday life of the inhabitants and can lead to significant economic loss. Hence, routine inspections are essential to sustain the water supply among the communities. Nowadays, inspection platforms with cameras, which can record the internal condition of the pipelines emerge to be one of the most attractive solutions. However, these inspection platforms often encounter some problems regarding autonomous navigation. As a result, an interruption occurs in the videotaping process which hampers the condition assessment process for the pipelines. Therefore, this paper presents a deep learning-based environment prediction model, which can predict the next instance of the environment inside the pipelines and enhance the autonomous navigation of the pipelines. The main objective of this paper is to make the inspection platforms intelligent enough to activate the control mechanisms and to pass through branches, curvature, elbows, etc. The results from the study show that integration of the environment prediction model with an embedded device can enhance the autonomous navigation inside the pipelines. This will also aid in the uninterrupted videotaping process and ensure better condition assessment of the water pipelines.