4D Radar-Inertial Odometry based on Gaussian Modeling and Multi-Hypothesis Scan Matching

2024-12-18

arXiv.org

AI Summary

Traditional point cloud–based odometry fails under adverse weather conditions (e.g., rain, snow, fog) due to sparse and noisy point clouds from 4D millimeter-wave radar. To address this, we propose a tightly coupled radar–inertial SLAM framework integrating Gaussian probabilistic representation with multi-hypothesis optimization. Specifically, we model radar point clouds using free-form 3D Gaussians to formulate a probabilistic scan-matching approach analogous to the Normal Distributions Transform (NDT). A parallel multi-hypothesis optimization mechanism is designed to enhance convergence robustness. Furthermore, we establish a tightly coupled fusion architecture leveraging an Extended Kalman Filter (EKF) for joint radar and IMU state estimation. Extensive experiments on mainstream 4D radar datasets demonstrate that our method significantly outperforms existing baselines, achieving high-accuracy and robust localization even under low signal-to-noise ratio (SNR) conditions.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

4D millimeter-wave (mmWave) radars are sensors that provide robustness against adverse weather conditions (rain, snow, fog, etc.), and as such they are increasingly being used for odometry and SLAM applications. However, the noisy and sparse nature of the returned scan data proves to be a challenging obstacle for existing point cloud matching based solutions, especially those originally intended for more accurate sensors such as LiDAR. Inspired by visual odometry research around 3D Gaussian Splatting, in this paper we propose using freely positioned 3D Gaussians to create a summarized representation of a radar point cloud tolerant to sensor noise, and subsequently leverage its inherent probability distribution function for registration (similar to NDT). Moreover, we propose simultaneously optimizing multiple scan matching hypotheses in order to further increase the robustness of the system against local optima of the function. Finally, we fuse our Gaussian modeling and scan matching algorithms into an EKF radar-inertial odometry system designed after current best practices. Experiments show that our Gaussian-based odometry is able to outperform current baselines on a well-known 4D radar dataset used for evaluation.

Problem

• Overcoming noise and sparsity in 4D radar data for odometry
• Enhancing robustness against local optima in scan matching
• Integrating radar-inertial odometry with Gaussian modeling techniques

Innovation

• Uses 3D Gaussians for radar point cloud representation
• Optimizes multiple scan matching hypotheses simultaneously
• Integrates Gaussian modeling into EKF radar-inertial odometry