CORENet: Cross-Modal 4D Radar Denoising Network with LiDAR Supervision for Autonomous Driving

2025-08-18

AI Summary

4D radar point clouds suffer from extreme sparsity and high noise levels, severely degrading perception robustness for autonomous driving under adverse weather conditions. To address this, we propose a LiDAR-guided cross-modal denoising framework: during training, high-fidelity LiDAR point clouds serve as supervision signals, while at inference, the model operates end-to-end on radar-only input—enabling plug-and-play deployment. We design a cross-modal neural network that jointly models radar noise distributions and LiDAR geometric priors, thereby enhancing noise pattern discrimination and feature reconstruction. Evaluated on the high-noise Dual-Radar dataset, our method significantly outperforms state-of-the-art denoising and detection baselines, achieving an 8.2% mAP improvement in heavy rain and dense fog scenarios. Moreover, it seamlessly integrates with existing voxel-based 3D detection pipelines without architectural modification, establishing a new paradigm for robust radar perception.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

4D radar-based object detection has garnered great attention for its robustness in adverse weather conditions and capacity to deliver rich spatial information across diverse driving scenarios. Nevertheless, the sparse and noisy nature of 4D radar point clouds poses substantial challenges for effective perception. To address the limitation, we present CORENet, a novel cross-modal denoising framework that leverages LiDAR supervision to identify noise patterns and extract discriminative features from raw 4D radar data. Designed as a plug-and-play architecture, our solution enables seamless integration into voxel-based detection frameworks without modifying existing pipelines. Notably, the proposed method only utilizes LiDAR data for cross-modal supervision during training while maintaining full radar-only operation during inference. Extensive evaluation on the challenging Dual-Radar dataset, which is characterized by elevated noise level, demonstrates the effectiveness of our framework in enhancing detection robustness. Comprehensive experiments validate that CORENet achieves superior performance compared to existing mainstream approaches.

Problem

• Denoise sparse 4D radar point clouds
• Improve radar perception robustness
• Enable radar-only inference with LiDAR supervision

Innovation

• Cross-modal denoising with LiDAR supervision
• Plug-and-play voxel integration architecture
• Training supervision without inference dependency