A Simple yet Effective Subway Self-positioning Method based on Aerial-view Sleeper Detection

TL;DR

A low-cost and real-time visual-assisted self-localization framework to address the robust and convenient positioning problem for subways and achieves a mean percentage error (MPE) of 0.1%, demonstrating its continuous and high-precision self-localization capability.

Abstract

With the rapid development of urban underground rail vehicles,subway positioning, which plays a fundamental role in the traffic navigation and collision avoidance systems, has become a research hot-spot these years. Most current subway positioning methods rely on localization beacons densely pre-installed alongside the railway tracks, requiring massive costs for infrastructure and maintenance, while commonly lacking flexibility and anti-interference ability. In this paper, we propose a low-cost and real-time visual-assisted self-localization framework to address the robust and convenient positioning problem for subways. Firstly, we perform aerial view rail sleeper detection based on the fast and efficient YOLOv8n network. The detection results are then used to achieve real-time correction of mileage values combined with geometric positioning information, obtaining precise subway locations. Front camera Videos for subway driving scenes along a 6.9 km route are collected and annotated from the simulator for validation of the proposed method. Experimental results show that our aerial view sleeper detection algorithm can efficiently detect sleeper positions with F1-score of 0.929 at 1111 fps, and that the proposed positioning framework achieves a mean percentage error of 0.1\%, demonstrating its continuous and high-precision self-localization capability.

Figures (8)

• Figure 1: Framework of the proposed subway self-positioning system. The entire framework takes in real-time inputs of the forward camera and speed sensor. Aerial view rail sleeper detection module performs visual transformation on the input image and detects the positions of sleepers on the railway track. Real-time localization estimation module utilizes the visual correction factor to periodically correct the accumulating errors in the mileage. Ultimately, the framework outputs the precise position of the subway train on its route.
• Figure 2: Image perspective conversion process (IPM). Illustration of the mapping relationships and the transformation process among real-world coordinate system ($O-XYZ$), front view image plane ($X_f-O_f-Y_f$), and aerial view image plane ($X_a-O_a-Y_a$).
• Figure 3: Illustration of the real-time localization estimation process. In adjacent moments $t_{1}$ and $t_{2}$, sleepers captured by the front camera are detected in the aerial view (Camera View), where the nearest sleeper provides a visual correction factor to estimate the exact advance distance for a subway train (Real World View).
• Figure 4: Illustration of the subway route and video frames (from front camera). The total route is about 6.9 km with color gray for the tunnels and color green for non-tunnel environments. Two typical example videos for the train running in and outside of subway tunnels are shown below, colored in correspondence with the route.
• Figure 5: Examples for aerial view images of rail sleepers. (a) and (b) show cases of different lighting and road conditions respectively.