Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Conceptual Evaluation of Deep Visual Stereo Odometry for the MARWIN Radiation Monitoring Robot in Accelerator Tunnels

André Dehne, Juri Zach, Peer Stelldinger

TL;DR

The paper tackles autonomous navigation for MARWIN in GPS-denied accelerator tunnels, where drift and infrastructural rigidity hinder existing QR-code-based localization. It conceptually evaluates deep visual stereo odometry (DVSO) against wheel and 2D LiDAR odometry, using a graph-based framework with PuzzlePole landmarks for ground-truth validation. Findings show DVSO offers translational accuracy similar to wheel odometry but incurs higher rotational errors and substantial computational overhead, while 2D LiDAR underperforms in monotone tunnel sections; landmark-driven graph optimization can greatly reduce drift. The work highlights a pragmatic design insight: rely on wheel odometry for translation and use LiDAR or landmarks for rotation correction, while advancing self-supervised landmark discovery and embedded hardware optimization for future, more autonomous navigation in constrained, safety-critical tunnels.

Abstract

The MARWIN robot operates at the European XFEL to perform autonomous radiation monitoring in long, monotonous accelerator tunnels where conventional localization approaches struggle. Its current navigation concept combines lidar-based edge detection, wheel/lidar odometry with periodic QR-code referencing, and fuzzy control of wall distance, rotation, and longitudinal position. While robust in predefined sections, this design lacks flexibility for unknown geometries and obstacles. This paper explores deep visual stereo odometry (DVSO) with 3D-geometric constraints as a focused alternative. DVSO is purely vision-based, leveraging stereo disparity, optical flow, and self-supervised learning to jointly estimate depth and ego-motion without labeled data. For global consistency, DVSO can subsequently be fused with absolute references (e.g., landmarks) or other sensors. We provide a conceptual evaluation for accelerator tunnel environments, using the European XFEL as a case study. Expected benefits include reduced scale drift via stereo, low-cost sensing, and scalable data collection, while challenges remain in low-texture surfaces, lighting variability, computational load, and robustness under radiation. The paper defines a research agenda toward enabling MARWIN to navigate more autonomously in constrained, safety-critical infrastructures.

Conceptual Evaluation of Deep Visual Stereo Odometry for the MARWIN Radiation Monitoring Robot in Accelerator Tunnels

TL;DR

The paper tackles autonomous navigation for MARWIN in GPS-denied accelerator tunnels, where drift and infrastructural rigidity hinder existing QR-code-based localization. It conceptually evaluates deep visual stereo odometry (DVSO) against wheel and 2D LiDAR odometry, using a graph-based framework with PuzzlePole landmarks for ground-truth validation. Findings show DVSO offers translational accuracy similar to wheel odometry but incurs higher rotational errors and substantial computational overhead, while 2D LiDAR underperforms in monotone tunnel sections; landmark-driven graph optimization can greatly reduce drift. The work highlights a pragmatic design insight: rely on wheel odometry for translation and use LiDAR or landmarks for rotation correction, while advancing self-supervised landmark discovery and embedded hardware optimization for future, more autonomous navigation in constrained, safety-critical tunnels.

Abstract

The MARWIN robot operates at the European XFEL to perform autonomous radiation monitoring in long, monotonous accelerator tunnels where conventional localization approaches struggle. Its current navigation concept combines lidar-based edge detection, wheel/lidar odometry with periodic QR-code referencing, and fuzzy control of wall distance, rotation, and longitudinal position. While robust in predefined sections, this design lacks flexibility for unknown geometries and obstacles. This paper explores deep visual stereo odometry (DVSO) with 3D-geometric constraints as a focused alternative. DVSO is purely vision-based, leveraging stereo disparity, optical flow, and self-supervised learning to jointly estimate depth and ego-motion without labeled data. For global consistency, DVSO can subsequently be fused with absolute references (e.g., landmarks) or other sensors. We provide a conceptual evaluation for accelerator tunnel environments, using the European XFEL as a case study. Expected benefits include reduced scale drift via stereo, low-cost sensing, and scalable data collection, while challenges remain in low-texture surfaces, lighting variability, computational load, and robustness under radiation. The paper defines a research agenda toward enabling MARWIN to navigate more autonomously in constrained, safety-critical infrastructures.

Paper Structure

This paper contains 22 sections, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Challenges in Accelerator Tunnels
  3. State of the Art
  4. Analysis by Focus Area
  5. FA1: Self-Supervised SLAM in Tunnels
  6. FA2: Resource-Efficient SLAM
  7. FA3: Self-Supervised Landmark Discovery
  8. Identified Research Gap
  9. Research Questions
  10. Contribution and Scope of this Work
  11. Deep Visual Stereo Odometry (DVSO)
  12. Experimental Setup
  13. Test Environment
  14. Robot Platform and Sensor Configuration
  15. Trajectory and Data Collection
  16. ...and 7 more sections

Figures (4)

  • Figure 1: Monotonous tunnel section at European XFEL used for odometry evaluation
  • Figure 2: Mobile robot platform
  • Figure 3: DVSO trajectory comparison: raw estimates (left) showing downward/rotational drift, and graph-optimized trajectory (right)
  • Figure 4: Wheel odometry trajectory comparison: raw measurements (left) showing cumulative drift, and graph-optimized trajectory (right)