Keywords: trams, driverless, driving path, rail detection, deep learning, boundary.

In this research, the tram driving path identification system for driverless trams is proposed. The primary purpose of the system is to identify the path in front of a tram where the tram will occupy. With this system, it is highly expected that it will substantially assist with preventing collisions between trams and frontal obstacles, and other relevant hazardous incidents. For the proposed system, it consists of two main modules: the rail detection module and the boundary generating module. In the study of rail detection module, the comparative analysis of different types of rail detectors, including traditional methods and learning-based methods, is performed to investigate the most suitable rail detectors. The comparative results reveal that the deep learning-based models are more accurate and reliable. Further, the boundary generating module is designed. Subsequently, the preliminary test of the proposed identification system based on the rail detection and designed boundary generating modules is carried out. According to the preliminary result, the proposed identification system successfully managed to perform its designated task very well, hence, it can be said that this proposed tram driving path identification system is promising for driverless tram operations.

download the full-text of this paper (PDF, 2356 Kb)

go to the previous paper
go to the next paper
return to the table of contents
return to the volume description