[Paper Review] Line-based Road Structure Mapping Using Multi-beam LiDAR.

Xudong He, Junqiao Zhao+5 more|arXiv (Cornell University)|Apr 19, 2018

Remote Sensing and LiDAR Applications3 citations

TL;DR

This paper proposes a line-based SLAM method using multi-beam LiDAR to map road structures with high precision and efficiency by representing roads as polylines instead of grids or point clouds. The approach uses local probabilistic fusion for road boundary extraction and efficient line-based matching, achieving an average absolute matching error of 0.07m and relative error of 8.64% across three real-world scenes.

ABSTRACT

In this paper, we studied a line-based SLAM method for road structure mapping using multi-beam LiDAR. We propose to use the polyline as the basic mapping element instead of grid cell or point cloud, because the line-based representation is precise and lightweight, and it can directly generate vector-based HD map as demanded by autonomous driving systems. We explored: 1) The extraction and vectorization of road structures based on local probabilistic fusion. 2) The efficient line-based matching between frames of vectorized road structures. A specified road structure, the road boundary, is taken as an example. The results testified the feasibility and effectiveness of the proposed method. We applied our proposed mapping system in three different scenes and achieved the average absolute matching error of 0.07m, the average relative matching error of 8.64%.

Motivation & Objective

• To develop a lightweight and precise road structure mapping method suitable for autonomous driving systems.
• To address the limitations of grid-based or point cloud-based representations by adopting polyline-based vector mapping.
• To improve matching accuracy and computational efficiency in dynamic road environments using line-based features.
• To demonstrate feasibility and effectiveness of line-based representation in real-world road scenes.

Proposed method

• Utilizes multi-beam LiDAR data to extract road boundaries through local probabilistic fusion for robust feature detection.
• Converts detected road structures into vectorized polylines to enable efficient, precise, and lightweight representation.
• Employs a line-based matching algorithm to align polylines across consecutive LiDAR frames using geometric and topological constraints.
• Applies probabilistic fusion techniques to enhance the reliability of line segment extraction in noisy or complex environments.
• Optimizes the mapping pipeline for real-time performance while maintaining high accuracy in vector map generation.
• Validates the system on three distinct road scenes to assess robustness and generalization capability.

Experimental results

Research questions

• RQ1 Can polyline-based representation outperform traditional grid or point cloud methods in terms of precision and computational efficiency for road structure mapping?
• RQ2 How effectively can local probabilistic fusion extract and vectorize road boundaries from multi-beam LiDAR data?
• RQ3 What is the achievable accuracy of line-based matching between consecutive LiDAR frames in diverse road environments?
• RQ4 To what extent does the proposed method generalize across different road scenes with varying complexity and dynamics?

Key findings

• The proposed line-based SLAM method achieved an average absolute matching error of 0.07 meters across three real-world scenes.
• The system demonstrated a relative matching error of 8.64%, indicating strong consistency in trajectory estimation.
• The polyline-based representation enabled efficient and precise vector map generation, suitable for autonomous driving applications.
• Local probabilistic fusion effectively enhanced the robustness of road boundary extraction in complex environments.
• The line-based matching algorithm significantly improved computational efficiency compared to point cloud or grid-based approaches.
• The method proved feasible and effective in diverse road scenarios, confirming its potential for real-world deployment.