The Finer Points: A Systematic Comparison of Point-Cloud Extractors for Radar Odometry

2024-09-18

arXiv.org

AI Summary

Frequency-modulated continuous-wave (FMCW) radar point cloud extraction significantly impacts the performance of iterative closest point (ICP)-based radar odometry, yet no systematic evaluation exists to identify optimal extraction strategies. Method: This work conducts the first comprehensive benchmark of 13 radar point cloud extractors on real-world road data spanning 176 km, quantitatively assessing their accuracy, computational efficiency, and robustness in ICP-based odometry. Hyperparameter tuning is performed for each extractor. Contribution/Results: The K-strongest extractor achieves the highest localization accuracy with minimal computational overhead: its absolute trajectory error (ATE) improves by 13.59% and 24.94% over the mean ATE of all 13 methods on two benchmark datasets. Further hyperparameter optimization yields substantial additional accuracy gains. These findings establish a reproducible, empirically grounded framework for selecting and optimizing point cloud extractors—enabling lightweight, high-performance radar SLAM systems.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Calibration & Uncertainty Quantification
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Security and Privacy: Large-scale security measurements

Abstract

A key element of many odometry pipelines using spinning frequency-modulated continuous-wave (FMCW) radar is the extraction of a point-cloud from the raw signal. This extraction greatly impacts the overall performance of point-cloud-based odometry. This paper provides a first-of-its-kind, comprehensive comparison of 13 common radar point-cloud extractors for the task of iterative closest point based odometry in autonomous driving environments. Each extractor's parameters are tuned and tested on two FMCW radar datasets using approximately 176km of data from public roads. We find that the simplest, and fastest extractor, K-strongest, is the best overall extractor, consistently outperforming the average by 13.59% and 24.94% on each dataset, respectively. Additionally, we highlight the significance of tuning an extractor and the substantial improvement in odometry accuracy that it yields.

Problem

• Comparing 13 radar point-cloud extractors for odometry
• Evaluating extractor impact on autonomous driving performance
• Identifying K-strongest as the best-performing extractor

Innovation

• Compares 13 radar point-cloud extractors
• Uses K-strongest as best extractor
• Emphasizes tuning for odometry accuracy