• Ubuntu 20.04
• ROS Noetic
• Conda with Python 3.8
• CUDA >= 11.3 capable GPU.
• ImageMagick

1. Install conda environment with conda env create -f env.yaml.
2. Source environment conda activate radarize_ae and then pip install -e ..
3. Install cartographer_ros. Inside cartographer/, configuration_files/ and launch/ into <catkin_ws>/install_isolated/share/cartographer_ros/.
4. Source conda environment (if not already) and cartographer_ros environment.

conda activate radarize_ae
source <catkin_ws>/install_isolated/setup.bash

1. Download the dataset dataset.zip from the link and unzip into this directory.
2. Download the saved models outputs eval.zip from link and unzip into this directory.

To generate results in the paper, run the top-level script

./run_eval.sh 

Then,

1. Run ./slam_eval.sh to get the SLAM metrics.
2. Run ./odom_eval.sh to get the odometry metrics.

./run.sh