<div align ="center"> <!-- <img src="./assets/logo.png" width="20%"> --> <h3> IROS 2024: Multi-Robot Active Graph Exploration with Reduced Pose-SLAM Uncertainty via Submodular Optimization </h3>

Ruofei Bai^1,2, Shenghai Yuan¹, Hongliang Guo², Pengyu Yin¹, Wei-Yun Yau², Lihua Xie¹

¹ Nanyang Technological University, ² Institute for Infocomm Research (I2R), Agency for Science, Technology and Research (A*STAR)

<a href="https://ieeexplore.ieee.org/abstract/document/10802691"><img alt="Paper" src="https://img.shields.io/badge/Paper-IEEE%20Xplore-pink"/></a> <a href="https://arxiv.org/abs/2407.01013"><img alt="Paper" src="https://img.shields.io/badge/Paper-arXiv-8A2BE2"/></a>

<!-- <a href='https://drive.google.com/drive/folders/1UmZ3vA1cOunB-2wgz8T1fJDebhb-gmax?usp=sharing'><img src='https://img.shields.io/badge/Dataset-UMAD-green' alt='Code&Datasets'></a> <a href="https://www.youtube.com/watch?v=xORb4H-AyNw"><img alt="Video" src="https://img.shields.io/badge/Video-Youtube-red"/></a> <a href="https://github.com/IMRL/UMAD/blob/main/Doc/UMAD-Poster.pdf"><img alt="Poster" src="https://img.shields.io/badge/Poster-blue"/></a> --> </div>

CGE

This repo implements a SLAM-Aware Collaborative Graph Exploration (CGE) method, which finds quick coverage path for multiple robots, while forming a well-connected collaborative pose graph to reduce SLAM uncertainty. Approximation algorithms in submodular maximization are adopted to provided performance guarantees for the actively selected loop-closing actions (loop closures).

This work extends our previous work on single-robot SLAM-aware exploration to the multi-robot case. Follow this IEEE RA-L paper and open-sourced code for more details.

News

Our paper has been accpeted by IEEE/RSJ IROS 2024 !!!

Please follow this link to the Arxiv version. Please consider citing our paper if you find it helpful.

@inproceedings{bai2024multi,
  title={Multi-Robot Active Graph Exploration with Reduced Pose-SLAM Uncertainty via Submodular Optimization},
  author={Bai, Ruofei and Yuan, Shenghai and Guo, Hongliang and Yin, Pengyu and Yau, Wei-Yun and Xie, Lihua},
  booktitle={2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  pages={10229--10236},
  year={2024},
  organization={IEEE}
}

Requirements

1. Install python libraries networkx, scipy, statistics, pickle, pyyaml. They can be installed by using pip install xxx.

2. Install OR-Tools for python: python -m pip install ortools.

Usage

1. Specify save path in config.yaml
2. Run main.py
3. Visualize the results by running simulation.py. The code will read results from paths specified in config.yaml.

Results

Following are the robot's trajectories with (right) & without (left) active loop-closings.

<figure> <img src="./image/2robot.gif" alt="Alt Text" width="800" height="400"> <!-- <figcaption style="text-align:center;">Active TSP-based Method</figcaption> --> </figure> </div> <figure> <img src="./image/3robot.gif" alt="Alt Text" width="800" height="400"> <!-- <figcaption style="text-align:center;">Active TSP-based Method</figcaption> --> </figure> </div> <figure> <img src="./image/5robot.gif" alt="Alt Text" width="800" height="400"> <!-- <figcaption style="text-align:center;">Active TSP-based Method</figcaption> --> </figure> </div>