Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Comparative Analysis of Visual Odometry in Virtual and Real-World Railways Environments

Gianluca D'Amico, Mauro Marinoni, Giorgio Buttazzo

TL;DR

The paper addresses the paucity of labeled railway data for perception tasks by proposing RailSim, an Unreal Engine 5–based synthetic data framework, to enable early-stage testing of visual odometry and SLAM. It evaluates the ORB-SLAM2 system in both monocular and stereo configurations on synthetic RailSim sequences and real OsDaR23 data, as well as KITTI highway sequences for benchmarking. Findings indicate monocular SLAM on RailSim closely matches real-world OsDaR23 performance, while stereo SLAM on RailSim approaches KITTI highway results, albeit with higher absolute errors in longer sequences; these differences are traced to scene richness and motion dynamics. The work demonstrates sim-to-real viability for railway perception tasks and points to future improvements in train-dynamics modeling and broader perception-task support within RailSim.

Abstract

Perception tasks play a crucial role in the development of automated operations and systems across multiple application fields. In the railway transportation domain, these tasks can improve the safety, reliability, and efficiency of various perations, including train localization, signal recognition, and track discrimination. However, collecting considerable and precisely labeled datasets for testing such novel algorithms poses extreme challenges in the railway environment due to the severe restrictions in accessing the infrastructures and the practical difficulties associated with properly equipping trains with the required sensors, such as cameras and LiDARs. The remarkable innovations of graphic engine tools offer new solutions to craft realistic synthetic datasets. To illustrate the advantages of employing graphic simulation for early-stage testing of perception tasks in the railway domain, this paper presents a comparative analysis of the performance of a SLAM algorithm applied both in a virtual synthetic environment and a real-world scenario. The analysis leverages virtual railway environments created with the latest version of Unreal Engine, facilitating data collection and allowing the examination of challenging scenarios, including low-visibility, dangerous operational modes, and complex environments. The results highlight the feasibility and potentiality of graphic simulation to advance perception tasks in the railway domain.

A Comparative Analysis of Visual Odometry in Virtual and Real-World Railways Environments

TL;DR

The paper addresses the paucity of labeled railway data for perception tasks by proposing RailSim, an Unreal Engine 5–based synthetic data framework, to enable early-stage testing of visual odometry and SLAM. It evaluates the ORB-SLAM2 system in both monocular and stereo configurations on synthetic RailSim sequences and real OsDaR23 data, as well as KITTI highway sequences for benchmarking. Findings indicate monocular SLAM on RailSim closely matches real-world OsDaR23 performance, while stereo SLAM on RailSim approaches KITTI highway results, albeit with higher absolute errors in longer sequences; these differences are traced to scene richness and motion dynamics. The work demonstrates sim-to-real viability for railway perception tasks and points to future improvements in train-dynamics modeling and broader perception-task support within RailSim.

Abstract

Perception tasks play a crucial role in the development of automated operations and systems across multiple application fields. In the railway transportation domain, these tasks can improve the safety, reliability, and efficiency of various perations, including train localization, signal recognition, and track discrimination. However, collecting considerable and precisely labeled datasets for testing such novel algorithms poses extreme challenges in the railway environment due to the severe restrictions in accessing the infrastructures and the practical difficulties associated with properly equipping trains with the required sensors, such as cameras and LiDARs. The remarkable innovations of graphic engine tools offer new solutions to craft realistic synthetic datasets. To illustrate the advantages of employing graphic simulation for early-stage testing of perception tasks in the railway domain, this paper presents a comparative analysis of the performance of a SLAM algorithm applied both in a virtual synthetic environment and a real-world scenario. The analysis leverages virtual railway environments created with the latest version of Unreal Engine, facilitating data collection and allowing the examination of challenging scenarios, including low-visibility, dangerous operational modes, and complex environments. The results highlight the feasibility and potentiality of graphic simulation to advance perception tasks in the railway domain.
Paper Structure (10 sections, 4 figures, 2 tables)

This paper contains 10 sections, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Proposed Approach
  4. Graphic Simulation for Railway Environment
  5. Algorithm for Simultaneous Localization and Mapping
  6. Experimental results
  7. Preliminary concepts
  8. Monocular SLAM compared with OSDaR23
  9. Other experiments with RailSim
  10. Conclusions

Figures (4)

  • Figure 1: Samples scenarios created in RailSim: a) urban environment including moving objects; b) farmland/rural areas including agricultural activities; c) bridge; d) detail of the electrified line.
  • Figure 2: First image of the OSDaR23 "3_fire_site_3.1" sequence (left) and its virtual replica generated by Railsim (right).
  • Figure 3: Images taken from train perspective of: "5_station_bergedorf_5.1" OSDaR23 sequence (a); open landscape scenario of "RailSim_00" (b); "6_station_klein_flottbek_6.2" OSDaR23 sequence (c); open landscape scenario of "RailSim_01" (d).
  • Figure 4: Images taken from vheicle perspective of: the "KITTI_01" sequence (a); "RailSim_U" synthetic sequence (b).