MGPRL

This repository contains the ROS package that implements the works from the algorithm "MGPRL: Distributed Multi-Gaussian Processes for Wi-Fi-based Multi-Robot Relative Localization in Large Indoor Environments". The following figure shows the conceptual overview of MGPRL:

Experimental Demonstration and Video

MGPRL.mp4

Requirements & Environment Setup

The package has been tested on ROS Noetic for both simulated and hardware Turtlebot3 robot (waffle) & Turtlebot2 (2e and 2i). The following requirements are needed before installing the package:

1. ROS Installation: You should have installed a ROS distribution (Noetic) on Ubuntu 20.04.

2. Workspace Creation: Make sure you have created a ROS workspace. You can follow the ROS tutorial to do this.

3. **Install the rtabmap package with the following command:

   sudo apt-get install ros-noetic-rtabmap-ros

4. Install ROS Navigation Stack: You can install the ROS navigation stack using the following command:

   sudo apt-get install ros-noetic-navigation

5. Python Modules Installation: Make sure python version is 3.8+. Install the following Python modules using pip:

   pip install numpy pandas matplotlib open3d opencv-python scikit-learn

6. Turtlebot3 Installation:

   cd ~/catkin_ws/src/
   git clone https://github.com/ROBOTIS-GIT/turtlebot3_simulations.git
   echo "export TURTLEBOT3_MODEL=waffle" >> ~/.bashrc
   source ~/.bashrc
   sudo apt-get install ros-noetic-turtlebot3*

7. Exploration Framework Installation: We utilize SPACE for exploration in ROS simulations.

   cd ~/catkin_ws/src/
   git clone https://github.com/herolab-uga/SPACE-MAP.git

8. AWS Gazebo Worlds Installation:

   cd ~/catkin_ws/src/
   https://github.com/aws-robotics/aws-robomaker-small-house-world.git
   https://github.com/aws-robotics/aws-robomaker-bookstore-world.git

9. Move_Base Navigation:

   sudo apt-get install ros-$ROS_noetic-move-base

10. ** Python Libraries Installation**:

    pip install --upgrade pip
    # Install required Python libraries
    pip install opencv-python
    pip install matplotlib
    pip install scikit-learn
    pip install scipy
    pip install numpy
    pip install pandas
    pip install GPy

11. Environment Initialization:

    cd ~/catkin_ws/src/
    git clone https://github.com/herolab-uga/SPACE-MAP.git
    catkin_make
    source devel/setup.bash

ROS Nodes

• Gazebo Simulation: Launches simulation environments (e.g., AWS House, AWS Bookstore) using launch files.
• Robot State Publisher: Publishes robot states from URDF models at high frequency (e.g., 100 Hz).
• Model Spawner: Places TurtleBot3 models in Gazebo at specified coordinates.
• RTAB-Map SLAM: Processes RGB-D inputs for SLAM, operating at specified frequencies and saving mapping data.
• RGBD Synchronization: Synchronizes RGB and depth image streams for SLAM processing.
• RTAB-Map Visualization: Visualizes SLAM outputs and odometry for debugging and analysis.
• Map Merger: Merges grid maps from multiple robots even with unknown initial positions.
• Move Base: Manages local and global path planning and navigation.
• Frontier Assigner: Allocates exploration tasks to robots based on predefined criteria.
• RSSI Simulation Node: RSSI_Simulation_House.py – simulates RSSI fields with multiple fixed Access Points.
• RSSI Robot Nodes:
  • rssi_robot_1.py
  • rssi_robot_2.py
  • rssi_robot_3.py
    These nodes collect RSSI measurements (using the robot’s ground-truth pose), perform multi-output Gaussian Process regression to predict RSSI fields, detect global and local maxima with uncertainty-based weighting, compute convex hulls of detected AP positions, and publish their AP estimates (in their own frame) while subscribing to similar data from other robots for relative localization.

ROS Topics

• Odometry & Control
  • ${robot_namespace}/odom: Provides odometry data.
  • ${robot_namespace}/cmd_vel: Receives velocity command inputs.
• Mapping & Perception
  • ${robot_namespace}/cloud_map: Outputs individual robot maps (e.g., from RTAB-Map).
  • ${robot_namespace}/scan: Publishes laser scan data for navigation.
  • ${robot_namespace}/camera/rgb/image_raw & ${robot_namespace}/camera/depth/image_raw: Publish RGB and depth image streams.
  • ${robot_namespace}/camera/rgb/camera_info: Provides metadata for the RGB camera.
  • ${robot_namespace}/move_base: Each robot’s Move Base server for local/global planning.
• Map Merging & Frontier Detection
  • /total_map: Publishes the merged 2D grid map.
  • /frontier_marker: Publishes 2D/3D spatial frontier markers.
  • /{target_name}/masker/masked_image_raw: Publishes masked images for feature extraction and mapping.
• RSSI & Relative Localization
  • /robot1_pose, /robot2_pose, /robot3_pose: Publish each robot’s ground-truth pose.
  • /robot1_ap_data, /robot2_ap_data, /robot3_ap_data: Each robot publishes its detected AP positions (global and candidate positions) with uncertainty weights in its own frame.
  • /robot1_relative, /robot2_relative, /robot3_relative: Each robot publishes computed relative transform data (alignment results) for other robots.

Launch

Run one of the example launch (includes rviz) files included in this ROS package. For instance,

    roslaunch <package_name> house.launch/ bookstore.launch

Contributions

• Sai Krishna Ghanta - PhD Candidate
• Dr. Ramviyas Parasuraman - Lab Director

Heterogeneous Robotics Lab

School of Computing, University of Georgia.

For further information, please contact Dr. Ramviyas Parasuraman at ramviyas@uga.edu.