On-track Localisation based on Object Detection
Title: On-track Localisation based on Object Detection
Authors: Masoumeh Rahimi (Cranfield University); Miftahur Rahman (Cranfield University); Isidro Durazo Cardenas (Cranfield University); Andrew Starr (Cranfield University); Amanda Hall (Network Rail); Robert Anderson (Network Rail);
Year: 2022
Citation: Rahimi, M., Rahman, M., Durazo Cardenas, I., Starr, A., Hall, A., Anderson, R., (2022). On-track Localisation based on Object Detection. UKRAS22 Conference “Robotics for Unconstrained Environments” Proceedings, 42-43. doi: 10.31256/Br7Bk8O
Abstract:
Abstract—Railway vehicle positioning is a task of fundamental importance in providing an accurate and reliable localisation information for railway control and safety operation. Acquiring the vehicle’s position in absolute level is also important to the maintenance sector, as it would lead the maintenance tasks to be done more accurate, and faster. In this
paper object detection based on feature matching technique is proposed to improve onboard train positioning systems to achieve higher levels of positioning performance. An RGB-D camera is integrated to the front-base of the vehicle for this purpose. This localisation technique examines the vehicle position on the railway track by extracting features from trackside infrastructure and comparing them with a predefined database.