Real-Time Dense Monocular SLAM with Neural Radiance Fields

Antoni Rosinol · John J. Leonard · Luca Carlone

Paper | Video |

Install

Clone repo with submodules:

git clone https://github.com/ToniRV/NeRF-SLAM.git --recurse-submodules
git submodule update --init --recursive

From this point on, use a virtual environment... Install torch (see here for other versions):

# CUDA 11.3
pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 --extra-index-url https://download.pytorch.org/whl/cu113

Pip install requirements:

pip install -r requirements.txt
pip install -r ./thirdparty/gtsam/python/requirements.txt

Compile ngp (you need cmake>3.22):

cmake ./thirdparty/instant-ngp -B build_ngp
cmake --build build_ngp --config RelWithDebInfo -j

Compile gtsam and enable the python wrapper:

cmake ./thirdparty/gtsam -DGTSAM_BUILD_PYTHON=1 -B build_gtsam 
cmake --build build_gtsam --config RelWithDebInfo -j
cd build_gtsam
make python-install

Install:

python setup.py install

Download Sample Data

This will just download one of the replica scenes:

./scripts/download_replica_sample.bash

Run

python ./examples/slam_demo.py --dataset_dir=./datasets/Replica/office0 --dataset_name=nerf --buffer=100 --slam --parallel_run --img_stride=2 --fusion='nerf' --multi_gpu --gui

This repo also implements Sigma-Fusion: just change --fusion='sigma' to run that.

FAQ

GPU Memory

This is a GPU memory intensive pipeline, to monitor your GPU usage, I'd recommend to use nvitop. Install nvitop in a local env:

pip3 install --upgrade nvitop

Keep it running on a terminal, and monitor GPU memory usage:

nvitop --monitor

If you consistently see "out-of-memory" errors, you may either need to change parameters or buy better GPUs :). The memory consuming parts of this pipeline are:

• Frame to frame correlation volumes (but can be avoided using on-the-fly correlation computation).
• Volumetric rendering (intrinsically memory intensive, tricks exist, but ultimately we need to move to light fields or some better representation (OpenVDB?)).

Installation issues

1. Gtsam not working: check that the python wrapper is installed, check instructions here: gtsam_python. Make sure you use our gtsam fork, which exposes more of gtsam's functionality to python.
2. Gtsam's dependency is not really needed, I just used to experiment adding IMU and/or stereo cameras, and have an easier interface to build factor-graphs. This didn't quite work though, because the network seemed to have a concept of scale, and it didn't quite work when updating poses/landmarks and then optical flow.
3. Somehow the parser converts this to const std::vector<const gtsam::Matrix&>&, and I need to remove manually in gtsam/build/python/linear.cpp the inner const X& ..., and also add <pybind11/stl.h> because:

  Did you forget to `#include <pybind11/stl.h>`?

Citation

@article{rosinol2022nerf,
  title={NeRF-SLAM: Real-Time Dense Monocular SLAM with Neural Radiance Fields},
  author={Rosinol, Antoni and Leonard, John J and Carlone, Luca},
  journal={arXiv preprint arXiv:2210.13641},
  year={2022}
}

License

This repo is BSD Licensed. It reimplements parts of Droid-SLAM (BSD Licensed). Our changes to instant-NGP (Nvidia License) are released in our fork of instant-ngp (branch feature/nerf_slam) and added here as a thirdparty dependency using git submodules.

Acknowledgments

This work has been possible thanks to the open-source code from Droid-SLAM and Instant-NGP, as well as the open-source datasets Replica and Cube-Diorama.

Contact

I have many ideas on how to improve this approach, but I just graduated so I won't have much time to do another PhD... If you are interested in building on top of this, feel free to reach out :) [email protected]