Our environment has been tested on 20.04 (CUDA 11.8).
Install ROS Noetic
following the instructions.
sudo apt install -y ros-noetic-rviz-imu-plugin
Initialize ROS Workspace.
mkdir -p ~/Workspace/active_inr_s_ws/src && cd ~/Workspace/active_inr_s_ws/src && catkin_init_workspace
Clone the repo and create conda environment
git clone [email protected]:kzj18/activeINR-S.git ~/Workspace/active_inr_s_ws/src/activeINR-S && cd ~/Workspace/active_inr_s_ws/src/activeINR-S
git submodule update --init --progress
conda env create -f environment.yml
conda activate activeINR_S
Install pytorch by following the instructions. For torch 2.0.1 with CUDA version 11.8:
pip install torch==2.0.1+cu118 torchvision==0.15.2+cu118 torchaudio==2.0.2+cu118 --extra-index-url https://download.pytorch.org/whl/cu118
# Ubuntu 20.04
pip install -r requirements.txt
pip install git+ssh://[email protected]/facebookresearch/pytorch3d.git
pip install git+ssh://[email protected]/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch
Habitat-lab and habitat-sim need to be installed for simulation. We use v0.2.3 (git checkout tags/v0.2.3
) for habitat-sim & habitat-lab and install the habitat-sim with the flag --with-cuda
.
cd ~/Workspace/active_inr_s_ws/src/activeINR-S/habitat/habitat-lab && git checkout tags/v0.2.3
pip install -e habitat-lab
pip install -e habitat-baselines
cd ~/Workspace/active_inr_s_ws/src/activeINR-S/habitat/habitat-sim && git checkout tags/v0.2.3
git submodule update --init --progress --recursive
python setup.py install --with-cuda
# For Ubuntu 20.04
cd ~/Workspace/active_inr_s_ws && catkin_make -DPYTHON_EXECUTABLE=/usr/bin/python3
echo "source ~/Workspace/active_inr_s_ws/devel/setup.bash" >> ~/.bashrc
Copy config template from config/.templates/user_config.json
to config/user_config.json
and modify the path to the dataset.
-
To run our method on the
Denmark
scene ofGibson
dataset, run the following command.roslaunch active_inr_s habitat.launch
-
To run our method on the
Pablo
scene ofGibson
dataset, run the following command.roslaunch active_inr_s habitat.launch scene_id:=Pablo
-
To run our method on the
zsNo4HB9uLZ
scene ofMP3D
dataset, run the following command.roslaunch active_inr_s habitat.launch config:=config/datasets/mp3d.json
-
To run our method on the
YmJkqBEsHnH
scene ofMP3D
dataset, run the following command.roslaunch active_inr_s habitat.launch config:=config/datasets/mp3d.json scene_id:=YmJkqBEsHnH
python scripts/entry_points/batch/iros_run.py
python scripts/entry_points/batch/eval_results_actions.py
@inproceedings{Kuang2024iros,
title={Active Neural Mapping at Scale},
author={Kuang, Zijia and Yan, Zike and Zhao, Hao and Zhou, Guyue and Zha, Hongbin},
booktitle={IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS)},
year={2024}
}