Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Evaluating Low-Resource Lane Following Algorithms for Compute-Constrained Automated Vehicles

Beñat Froemming-Aldanondo, Tatiana Rastoskueva, Michael Evans, Marcial Machado, Anna Vadella, Rickey Johnson, Luis Escamilla, Milan Jostes, Devson Butani, Ryan Kaddis, Chan-Jin Chung, Joshua Siegel

TL;DR

This work tackles reliable lane following on compute-constrained automated vehicles by evaluating five low-resource algorithms—Largest White Contour, Bird's-eye View with Least Squares, Linear Lane Search with K-Means, Lane Line DBSCAN, and DeepLSD—through simulation and real-drive tests on drive-by-wire Polaris vehicles. The methods are assessed across reliability, comfort, speed, and adaptability, with processing times under 10 ms per frame. Results show unsupervised clustering approaches (K-Means and DBSCAN) offer robust, real-time lane detection, while DeepLSD serves as a higher-cost benchmark that struggles with curved or noisy lines. The findings support deploying efficient lane-detection techniques to broaden access to lane keeping and lower-level automation on resource-limited platforms.

Abstract

Reliable lane-following is essential for automated and assisted driving, yet existing solutions often rely on models that require extensive computational resources, limiting their deployment in compute-constrained vehicles. We evaluate five low-resource lane-following algorithms designed for real-time operation on vehicles with limited computing resources. Performance was assessed through simulation and deployment on real drive-by-wire electric vehicles, with evaluation metrics including reliability, comfort, speed, and adaptability. The top-performing methods used unsupervised learning to detect and separate lane lines with processing time under 10 ms per frame, outperforming compute-intensive and poor generalizing deep learning approaches. These approaches demonstrated robustness across lighting conditions, road textures, and lane geometries. The findings highlight the potential for efficient lane detection approaches to enhance the accessibility and reliability of autonomous vehicle technologies. Reducing computing requirements enables lane keeping to be widely deployed in vehicles as part of lower-level automation, including active safety systems.

Evaluating Low-Resource Lane Following Algorithms for Compute-Constrained Automated Vehicles

TL;DR

This work tackles reliable lane following on compute-constrained automated vehicles by evaluating five low-resource algorithms—Largest White Contour, Bird's-eye View with Least Squares, Linear Lane Search with K-Means, Lane Line DBSCAN, and DeepLSD—through simulation and real-drive tests on drive-by-wire Polaris vehicles. The methods are assessed across reliability, comfort, speed, and adaptability, with processing times under 10 ms per frame. Results show unsupervised clustering approaches (K-Means and DBSCAN) offer robust, real-time lane detection, while DeepLSD serves as a higher-cost benchmark that struggles with curved or noisy lines. The findings support deploying efficient lane-detection techniques to broaden access to lane keeping and lower-level automation on resource-limited platforms.

Abstract

Reliable lane-following is essential for automated and assisted driving, yet existing solutions often rely on models that require extensive computational resources, limiting their deployment in compute-constrained vehicles. We evaluate five low-resource lane-following algorithms designed for real-time operation on vehicles with limited computing resources. Performance was assessed through simulation and deployment on real drive-by-wire electric vehicles, with evaluation metrics including reliability, comfort, speed, and adaptability. The top-performing methods used unsupervised learning to detect and separate lane lines with processing time under 10 ms per frame, outperforming compute-intensive and poor generalizing deep learning approaches. These approaches demonstrated robustness across lighting conditions, road textures, and lane geometries. The findings highlight the potential for efficient lane detection approaches to enhance the accessibility and reliability of autonomous vehicle technologies. Reducing computing requirements enables lane keeping to be widely deployed in vehicles as part of lower-level automation, including active safety systems.
Paper Structure (18 sections, 2 equations, 11 figures, 1 table)

This paper contains 18 sections, 2 equations, 11 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Traditional Methods
  4. AI Methods
  5. Material and Methods
  6. Simulation and Real Environment
  7. Vehicle Specifications
  8. Architecture and Dynamic Reconfigure
  9. Preprocessing
  10. Lane Detection Algorithms
  11. Largest White Contour
  12. Birds-eye-view with Least Square Regression line fitting
  13. Linear Lane Search with K-Means
  14. Lane Line classification using DBSCAN
  15. DeepLSD Lane Detection
  16. ...and 3 more sections

Figures (11)

  • Figure 1: Aerial view of the Lot H course in LTU.
  • Figure 2: ACTors 1 and 2.
  • Figure 3: Largest Contour and Offset Point.
  • Figure 4: Birds-eye-view transformation with fitted lane lines.
  • Figure 5: Visualization of Linear Lane Search with K-Means.
  • ...and 6 more figures