RMap: Millimeter-Wave Radar Mapping Through Volumetric Upsampling

TL;DR

This paper introduces RMap, a novel transformer-based method that upsamples and denoises millimeter-wave radar data to produce accurate 3D maps comparable to lidar, especially useful in adverse conditions.

Contribution

The paper presents UpPoinTr, a new generative transformer architecture designed to enhance sparse radar point clouds into detailed 3D maps, addressing noise and sparsity issues.

Findings

RMap effectively denoises and upsamples radar data

Produces lidar-like 3D maps from sparse radar point clouds

Demonstrates superior performance on ColoRadar dataset

Abstract

Millimeter Wave Radar is being adopted as a viable alternative to lidar and radar in adverse visually degraded conditions, such as the presence of fog and dust. However, this sensor modality suffers from severe sparsity and noise under nominal conditions, which makes it difficult to use in precise applications such as mapping. This work presents a novel solution to generate accurate 3D maps from sparse radar point clouds. RMap uses a custom generative transformer architecture, UpPoinTr, which upsamples, denoises, and fills the incomplete radar maps to resemble lidar maps. We test this method on the ColoRadar dataset to demonstrate its efficacy.