DGORL: Distributed Graph Optimization based Relative Localization of Multi-Robot Systems
By creating relative position-weighted connectivity graphs using RSSI as local sensor data, expanding these graphs based on potential positions at a particular location, and then further optimizing to obtain relative position estimates for all connected robots, DGORL seeks to efficiently achieve high localization accuracy. An overview of the DGORL configuration space can be found in Figure below:
Demonstration of DGROL implementation on real robots: Here, five swarm robots are executing a simple multi-robot rendezvous algorithm to group together. The left side shows the robot's overhead video, the center plot shows the actual trajectory based on AprilTag tracking, the right plot shows the DGROL's localization (with Graph) outcome.
If you use this work in your research, please cite it as Latif, E., Parasuraman, R. (2024). DGORL: Distributed Graph Optimization Based Relative Localization of Multi-robot Systems. In: Bourgeois, J., et al. Distributed Autonomous Robotic Systems. DARS 2022. Springer Proceedings in Advanced Robotics, vol 28. Springer, Cham. https://doi.org/10.1007/978-3-031-51497-5_18
Paper Link: https://link.springer.com/chapter/10.1007/978-3-031-51497-5_18
The preprint of the paper is available in ARXiv at https://arxiv.org/abs/2210.01662
- C++ requirements.
(pybind11 is also required, but it's built in this repository, you don't need to install) - python 3.6+
- g2o installed
$ git clone https://github.com/uoip/g2opy.git
$ cd g2opy
$ mkdir build
$ cd build
$ cmake ..
$ make -j8
$ cd ..
$ python setup.py install
Tested under Ubuntu 16.04, Python 3.6+.
This repository contains script graph_optimization_mrl.py to simulate the DGORL on 60 x 60 m workspace based on g2o optimizer along with the graph input files in g2o format.
$ git clone https://github.com/herolab-uga/DGORL.git
$ cd DGORL
$ python3 graph_optimization_mrl.py
You can set following paramters in script:
- number of robots
- number of iterations
- workspace dimensions
- number of max_iterations for optimizer
- dampting factory for optimizer
-
Ehsan Latif - PhD student
-
Dr. Ramviyas Parasuraman - Principal Investigator
Heterogeneous Robotics Lab (HeRoLab), Department of Computer Science, University of Georgia. http://hero.uga.edu
For further information, contact Ehsan Latif ehsan.latif@uga.edu or Prof. Ramviyas Parasuraman ramviyas@uga.edu
