Position Estimation of In-Pipe Robot Using Artificial Neural Network and Sensor Fusion

Abstract

Automatic position detection of water leakage in water distribution pipelines is critical to minimize the loss of labour, time, money spent on exploration and excavation in pipe inspection procedures. Nevertheless, the main goal of detection is to prevent water loss. In this paper, accurate position detection, crack frequency band detection, and external sphere studies of an in-pipe robot prototype were presented. During the precise position estimation, classical Extended Kalman Filter (EKF), stationary region detection, and location estimation using Enhanced Heuristic Drift Elimination (EHDE) were performed with two different artificial neural networks (ANNs). In this way, online precise position estimation can be done on hardware with no sufficient computational power for indoor robotic studies. In addition, the sound characteristics resulting from the crack at different hole size and water pressure intensity levels were investigated. Finally, a new sealing sphere design was devised. Three different hydrophone sensor data were recorded on the SD card simultaneously. The results show that the proposed ANN method can work online and make a similar position estimation with the classical IMU position estimation method by 99%. © 2021, Sakarya University. All rights reserved.