RadarSim: Simulating Single-Chip Radar via Multimodal Neural Fields

2026-05-25

AI Summary

This work addresses the challenges of using radar data for high-fidelity simulation and perception, which stem from its inherently low spatial resolution and sensor-specific discrepancies. To overcome these limitations, the authors propose RadarSim, a differentiable radar renderer that leverages the high angular resolution of RGB cameras by integrating camera-initialized multimodal neural radiance fields into radar simulation for the first time. The approach combines radar–camera joint calibration with a novel synchronized dataset collected via a custom handheld device, thereby transcending the constraints of conventional radar-only reconstruction methods. Experimental results demonstrate that RadarSim significantly enhances geometric detail recovery and Doppler-range image quality, outperforming existing purely radar-based reconstruction techniques.

Abstract

Radars are an ideal complement to cameras: both are inexpensive, solid-state sensors, with cameras offering fine angular resolution, while radars provide metric depth and robustness under adverse weather. However, radar data is more difficult to interpret than camera images and varies significantly between sensors, necessitating increased reliance on simulation for prototyping sensors and processing pipelines. Recent work treating radar reconstruction as a novel view synthesis problem has shown great promise in reconstructing radar-relevant geometry and simulating low-level radar data. However, such methods are constrained by the low spatial resolution of the underlying radar. To address this, we propose a unified differentiable renderer, RadarSim, which leverages the high angular resolution of RGB cameras to generate Doppler radar range images from a camera-initialized neural field. Using a novel data set of calibrated radar camera recordings from a custom hand-held rig, we demonstrate that RadarSim produces sharper geometry and Doppler range frames than radar-only reconstructions.

Problem

• radar simulation
• low spatial resolution
• multimodal sensing
• Doppler radar
• sensor fusion

Innovation

• multimodal neural fields
• differentiable rendering
• radar simulation
• camera-radar fusion
• Doppler radar range imaging