Radar Perception

A starting map for deep-learning-based radar perception.

Radar Perception

Radar perception sits at the intersection of sensing, geometry, representation learning, and deployment constraints. This page is a maintained starting point rather than a finished survey.

Starting points

• Task map: detection, tracking, segmentation, occupancy, and motion forecasting.
• Representation questions: point clouds, range-azimuth maps, voxelized features, and radar-camera fusion.
• Engineering questions: sparsity, multipath, calibration, temporal aggregation, and robustness in difficult weather.

Public index

The companion repository awesome-deeplearning-based-radar-perception collects papers, datasets, and implementations. Use it as the bibliography for this topic and open an issue when a link needs correction.

Next notes

This topic will grow through short, dated posts in the blog, then settle useful conclusions here.