Millimeter-wave radar offers a promising sensing modality for autonomous systems thanks to its robustness in adverse conditions and low cost. However, its utility is significantly limited by the sparsity and low resolution of radar point clouds, which poses challenges for tasks requiring dense and accurate 3D perception. Despite that recent efforts have shown great potential by exploring generative approaches to address this issue, they often rely on dense voxel representations that are inefficient and struggle to preserve structural detail. To fill this gap, we make the key observation that latent diffusion models (LDMs), though successful in other modalities, have not been effectively leveraged for radar-based 3D generation due to a lack of compatible representations and conditioning strategies. We introduce RaLD, a framework that bridges this gap by integrating scene-level frustum-based LiDAR autoencoding, order-invariant latent representations, and direct radar spectrum conditioning. These insights lead to a more compact and expressive generation process. Experiments show that RaLD produces dense and accurate 3D point clouds from raw radar spectrums, offering a promising solution for robust perception in challenging environments.

Automotive Millimeter-wave (mmWave) radar is becoming an essential modality for autonomous vehicles to enable all-weather perception, especially when LiDAR and camera fail in foggy, rainy, or snowy conditions. It is expected that the mutual interference among multiple radars becomes a critical issue in dense traffic scenarios, which can severely degrade the radar performance and lead to accidents. Despite extensive interference mitigation techniques, none can meet the less valid signal distortion while high robustness requirements for automotive radar perception in dense traffic scenarios. To overcome this predicament, we propose mmMic, a novel multiple mutual interference mitigation system that can accurately separate interference and recover valid signals to maintain the reliability of the radar measurements. The key insight is to design an interference estimator that can accurately localize the interference signal according to its linear frequency modulation features in the time-frequency (TF) domain. In addition, mmMic also fully exploits undisturbed valid signal information within an extended time-frequency domain to reconstruct the damaged signal. Our experiments on a real testbed show that mmMic can improve SINR to interference-free levels from multiple radars, achieving an average SINR improvement of 17% compared to the best-performing baseline.