Active Neural Mapping at Scale [IROS 2024]

Installation

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

Preparation

Simulated environment

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

Build activeINR-S

# 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

Run activeINR-S

Config Datasets Path

Copy config template from config/.templates/user_config.json to config/user_config.json and modify the path to the dataset.

Single Scene

1. To run our method on the Denmark scene of Gibson dataset, run the following command.

   roslaunch active_inr_s habitat.launch

2. To run our method on the Pablo scene of Gibson dataset, run the following command.

   roslaunch active_inr_s habitat.launch scene_id:=Pablo

3. To run our method on the zsNo4HB9uLZ scene of MP3D dataset, run the following command.

   roslaunch active_inr_s habitat.launch config:=config/datasets/mp3d.json

4. To run our method on the YmJkqBEsHnH scene of MP3D dataset, run the following command.

   roslaunch active_inr_s habitat.launch config:=config/datasets/mp3d.json scene_id:=YmJkqBEsHnH

Run IROS Results

python scripts/entry_points/batch/iros_run.py

Eval Results

python scripts/entry_points/batch/eval_results_actions.py

Citation

@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}
}