This repository contains implementation of synchronous and asynchronous Distributed Pose Graph Optimization (DPGO). The algorithms are described in the following publications:
-
Y. Tian, K. Khosoussi, D. M. Rosen, J. P. How. Distributed Certifiably Correct Pose-Graph Optimization, in IEEE Transactions on Robotics, 2021.
-
Y.Tian, A. Koppel, A. S. Bedi, J. P. How. Asynchronous and Parallel Distributed Pose Graph Optimization, in IEEE Robotics and Automation Letters, RA-L best paper honorable mention, 2020.
Install dependencies.
sudo apt-get install build-essential cmake-gui libsuitesparse-dev
Inside the C++ directory, execute the following commands.
mkdir build
cd build
cmake ../
make
The built executables are located in directory build/bin. For a serialized demo of distributed PGO on one of the benchmark datasets, inside the build directory run:
./bin/multi-robot-example 5 ../data/smallGrid3D.g2o
Optionally, run the unit tests by,
./bin/testDPGO
A ROS wrapper of dpgo is provided: dpgo_ros. If you want to use dpgo with ROS, please follow instructions in the README.
If you find this library useful, please consider citing our papers:
@ARTICLE{Tian2021Distributed,
author={Tian, Yulun and Khosoussi, Kasra and Rosen, David M. and How, Jonathan P.},
journal={IEEE Transactions on Robotics},
title={Distributed Certifiably Correct Pose-Graph Optimization},
year={2021},
volume={},
number={},
pages={1-20},
doi={10.1109/TRO.2021.3072346}}
@ARTICLE{Tian2020Asynchronous,
author={Tian, Yulun and Koppel, Alec and Bedi, Amrit Singh and How, Jonathan P.},
journal={IEEE Robotics and Automation Letters},
title={Asynchronous and Parallel Distributed Pose Graph Optimization},
year={2020},
volume={5},
number={4},
pages={5819-5826},
doi={10.1109/LRA.2020.3010216}}