Multi S-Graphs: an Efficient Real-time Distributed Semantic-Relational Collaborative SLAM

Miguel Fernandez Cortizas · Hriday Bavle · David Perez-Saura
Jose Luis Sanchez-Lopez · Pascual Campoy · Holger Voos

Paper | Video

TABLE OF CONTENTS

• Published Papers
• Getting started
• About Multi S-Graphs
  • Architecture
• Example on Datasets
  • Download Datasets
  • Run Multi S-Graphs
• Configuration
• ROS 2 Related
  • Relevant topics description
• License
• Maintainers

Published Papers

1. Multi S-Graphs: an Efficient Real-time Distributed Semantic-Relational Collaborative SLAM

   • Citation

     @misc{fernandezcortizas2024multi,
      title={Multi S-Graphs: an Efficient Real-time Distributed Semantic-Relational Collaborative SLAM}, 
      author={Miguel Fernandez-Cortizas and Hriday Bavle and David Perez-Saura and Jose Luis Sanchez-Lopez and Pascual Campoy and Holger Voos},
      year={2024},
      eprint={2401.05152},
      archivePrefix={arXiv},
      primaryClass={cs.RO}}

2. S-Graphs+: Real-time Localization and Mapping leveraging Hierarchical Representations

   • Citation

      @ARTICLE{10168233,
        author={Bavle, Hriday and Sanchez-Lopez, Jose Luis and Shaheer, Muhammad and Civera, Javier and Voos, Holger},
        journal={IEEE Robotics and Automation Letters}, 
        title={S-Graphs+: Real-Time Localization and Mapping Leveraging Hierarchical Representations}, 
        year={2023},
        volume={8},
        number={8},
        pages={4927-4934},
        doi={10.1109/LRA.2023.3290512}}

3. Situational Graphs for Robot Navigation in Structured Indoor Environments

   • Citation

      @ARTICLE{9826367,
        author={Bavle, Hriday and Sanchez-Lopez, Jose Luis and Shaheer, Muhammad and Civera, Javier and Voos, Holger},
        journal={IEEE Robotics and Automation Letters}, 
        title={Situational Graphs for Robot Navigation in Structured Indoor Environments}, 
        year={2022},
        volume={7},
        number={4},
        pages={9107-9114},
        doi={10.1109/LRA.2022.3189785}}

Getting started

1. Clone this repository

2. Pull the docker image from DockerHub

docker pull sntarg/multi_s_graphs:latest

3. Allow multicast for Zenoh and display access for visualization

sudo ifconfig lo multicast
xhost +

4. Modify the path/to/rosbags/folder inside the docker-compose.yml file.
5. Compose a container from image using the docker-compose.yml configuration.

cd [/path/to/repository/folder]
docker compose up -d

This command also incorporates the flags d, which makes the container run in the detached mode

6. Open a terminal inside the container

docker exec -ti ms_graphs bash

7. Execute an instance of Multi S-Graphs

./tmux_launch.bash [robot_namespace] [path/to/robot/rosbag]

About Multi S-Graphs

Architecture

Example on Datasets

Note: For each command below, please execute them in separate terminal windows!

Download Datasets

Real Dataset

Virtual Dataset

Run Multi S-Graphs

1. Open a terminal inside the container

docker exec -ti ms_graphs bash

2. In this terminal inside the container, execute an instance of Multi S-Graphs for Robot1

./tmux_launch.bash robot1 rosbags/?_split_robot1.bag

3. In a different window, open another terminal inside the container

docker exec -ti ms_graphs bash

4. In this terminal inside the container, execute an instance of Multi S-Graphs for Robot2

./tmux_launch.bash robot1 rosbags/?_split_robot2.bag

Configuration

The configuration of each robot is the same than described in S-Graphs+.

ROS 2 Related

The ROS 2 RMW that we use is CycloneDDS as we rely on the Zenoh ROS2 bridges for connecting multiple robots between them. Please check that your ROS 2 system is using this RMW if you want to inspect the execution nodes.

In our system we use a different ROS_DOMAIN_ID environment variable for each robot. The ID matches with the robot number e.g. robot1 has ROS_DOMAIN_ID=1. In order to inspect the different nodes and topic please set your terminal into a corresponding ROS_DOMAIN_ID with:

export ROS_DOMAIN_ID=<YOUR_DESIRED_ID>

Relevant topics description

Input topics for each robot are the same than described in S-Graphs+ with the appropiate namespaces.

Multi S-Graphs shared topics between agents:

• /graph_room_keyframes_all [graph_manager_msgs/msg/RoomKeyframe]
  • Shared Room Descriptors between agents.
• /graph_vertical_planes_all [graph_manager_msgs/msg/VerticalPlanes]
  • Shared Vertical Planes between agents.
• /relative_map_estimation [graph_manager_msgs/msg/RelativeMapEstimation]
  • Tranformations between every agent map frame.

Visualization topics for each robot:

• /[robot_namespace]/s_graphs/markers [visualization_msgs/msg/MarkerArray]
  • Visualization of the Situational Graph
• /[robot_namespace]/s_graphs/map_points_ts [sensor_msgs/msg/PointCloud2]
  • Visualization of the Point Cloud
• /[robot_namespace]/s_graphs/pose_ts [geometry_msgs/msg/PoseStamped]
  • Visualization of the Pose

License

This package is released under the BSD-2-Clause License.

Note that the cholmod solver in g2o is licensed under GPL. You may need to build g2o without cholmod dependency to avoid the GPL.

Maintainers

• Miguel Fernandez-Cortizas
  • Email: miguel.fernandez.cortizas@upm.es
• Hriday Bavle
  • Email: hriday.bavle@uni.lu
  • Website: https://www.hriday.bavle.com/
• David Perez-Saura
  • Email: david.perez.saura@upm.es