Abstract
Up-to-date geodatasets on railway infrastructure are valuable resources for the field of transportation. This paper investigates three methods for mapping the center lines of railway tracks using heterogeneous sensor data: (i) conditional selection of satellite navigation (GNSS) data, (ii) a combination of inertial measurements (IMU data) and GNSS data in a Kalman filtering and smoothing framework and (iii) extraction of center lines from laser scanner data. Several combinations of the methods are compared with a focus on mapping in tree-covered areas. The center lines of the railway tracks are extracted by applying these methods to a test dataset collected by a road-rail vehicle. The guard rails in the test area were also extracted during the center line detection process. The combination of methods (i) and (ii) gave the best result for the track on which the measurement vehicle had moved, mapping almost 100% of the track. The combination of methods (ii) and (iii) and the combination of all three methods gave the best result for the other parallel tracks, mapping between 25% and 80%. The mean perpendicular distance of the mapped center lines from the reference data was 1.49 meters.
Highlights
Geodatasets on railway infrastructure are useful resources in the transportation sector for routing applications and planning construction activities [1,2,3]
Research is driven towards algorithms that automatically map the railway infrastructure from data collected using a wide range of sensors such as digital cameras, laser scanners, global navigation satellite system (GNSS) receivers and eddy current sensors (ECS) [1,2,3,4,5,6,7,8]
The data collected by the GNSS receiver and Inertial measurement units (IMU) were given as input to a data fusion algorithm based on Kalman filtering and smoothing, which generated the trajectory of the vehicle that represents the center line of the driven track
Summary
Geodatasets on railway infrastructure are useful resources in the transportation sector for routing applications and planning construction activities [1,2,3]. Research is driven towards algorithms that automatically map the railway infrastructure from data collected using a wide range of sensors such as digital cameras, laser scanners, global navigation satellite system (GNSS) receivers and eddy current sensors (ECS) [1,2,3,4,5,6,7,8]. By following methods based on sensor data, both the cost and time are significantly reduced. This enables periodical repetition of the mapping process in order to account for changes in the infrastructure [2,4,5,6]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
