radarays_ros

RadaRays - ROS package

Rotating FMCW radar simulation based on ray-tracing. This package contains all the tools that has been used to simulate radar data and compare it to real sensor data.

Requirements

Mainly tested on:

And others: see paper (preprint is coming soon).

MulRan

Demonstration of simulating rotating FMCW radar data in large-scale triangle meshes from MulRan datasets. We reconstructed a mesh of the sequences DCC, and KAIST, and added a localization frame to each bag file using MICP-L as described here: https://github.com/aock/micp_experiments/tree/main/micp_mulran. The final bag files are piped to RadaRays. The simulation given the localization is started by

roslaunch radarays_ros mulran_sim.launch gui:=false

The simulated data is published on topic /radar/image. To enable the visualization you will need https://github.com/aock/mesh_tools. Then start the same launch file:

roslaunch radarays_ros mulran_sim.launch gui:=true

The results should look as follows:

BRDF

Left is the real radar polar image from the MulRan datasets. Right is the RadaRays simulated polar image.

Videos: https://youtube.com/playlist?list=PL9wBuzh6ev06rcl8ksSnxRtv-7jAkR_wx

BRDF

Easily customize the models reflection parameters using the following interactive tool:

rosrun radaray_ros radarays_snell_fresnel_brdf.py

BRDF

Citation

Please reference the following papers when using RadaRays in your scientific work.

Radarays Plugins

Roadmap

Development

There are currently the following active development tracks on individual branches:

  1. dev/flex
    • flexible reflection models for rapid prototyping (CPU-only)
    • Video for Cook-Torrance reflection model: https://www.youtube.com/watch?v=PI5j87NQzmk
  2. dev/opti
    • automatic material property optimization using the standard RadaRays simulation model

Individual components of these development branches are transferred to the main branch after extensive testing.