MR-APF: Multi-Robot Artificial Potential Fields Path Planning

A ROS-based implementation for distributed multi-robot path planning using Artificial Potential Fields (APF) for real-time navigation and obstacle avoidance. Features adaptive and multi-robot convergence avoidance potential fields for improved planning.

Overview

This project provides a multi-robot APF path planning system, enabling multiple robots to autonomously navigate while avoiding obstacles and each other. It supports both real-time Gazebo simulation and lightweight point-robot simulation with RViz visualization.

Key improvements include:

• Adaptive obstacle avoidance potentials
• Adaptive inter-robot collision avoidance
• New convergence avoidance potentials to prevent robot clustering

Features

• Distributed Planning: Each robot runs its own planner, sharing fleet data
• Real-time Visualization: Live monitoring in RViz
• Dual Simulation Modes: Choose between Gazebo (realistic physics) or point-robot simulation (fast, scalable)
• Dynamic Obstacle Avoidance: Handles moving obstacles and inter-robot collisions
• Fleet Coordination: Centralized data management with distributed execution
• Result Analysis: Automatic path recording and visualization

System Architecture

The system includes:

• Central MRAPF Service: Coordinates fleet data and robot initialization
• Robot Planners: Individual APF-based planners for each robot
• Fleet Data Handler: Manages robot positions and states
• Visualization System: Real-time RViz monitoring

Simulation Modes

Set the mode in the parameters file.

1. Real-time Gazebo Simulation

• Full physics simulation with TurtleBot3 models
• Realistic sensor data and robot dynamics

2. Simple Point-Robot Simulation

• Lightweight, fast simulation for algorithm testing
• Robots modeled as circular agents
• Scalable to large fleets

Both modes offer:

• Real-time RViz visualization
• Path recording and analysis
• Performance metrics collection

Sample Environment

Installation

1. Clone the repository into your ROS workspace:

cd ~/catkin_ws/src
git clone https://github.com/MortezaHagh/mr_apf.git

2. Build the package:

cd ~/catkin_ws
catkin_make
source devel/setup.bash

Usage

Quick Start

Wait a few seconds after runnign the nodes.

# For multi robot path planning in Gazebo simulation with TurtleBot3
roslaunch apf run_mrapf.launch

# for multi robot path planning - Simple Point-Robot Simulation
roslaunch apf point_sim.launch

Configuration

• Edit scripts/parameters.py for algorithm parameters
• Modify launch files for robot configurations
• Adjust map files in model_inputs.py

Project Structure

mr_apf/
├── action/          # ROS action definitions
├── design/          # Architecture diagrams and docs
├── launch/          # ROS launch files
├── maps/            # Environment maps (JSON, PNG, SVG)
├── models/          # Robot and obstacle models
├── msg/             # ROS message definitions
├── results/         # Simulation results and analysis
├── rviz/            # RViz configuration files
├── scripts/         # Core Python implementation
│   ├── apf_planner_*.py    # APF algorithm implementations
│   ├── mrapf_main.py       # Main simulation entry point
│   ├── visualization.py    # RViz visualization
│   └── ...
└── srv/             # ROS service definitions

Key Components

• APF Planners: Core artificial potential field algorithms
• Fleet Management: Centralized coordination system
• Visualization: Real-time RViz display
• Model Creation: Environment and robot model generation
• Results Analysis: Path and performance metrics

Results

Simulation results are saved in the results/ directory, including:

• Robot trajectories and paths
• Performance metrics
• Visualization plots
• JSON data for further analysis

License

This project is licensed under the MIT License. See the LICENSE file for details.

Citation

If you use this work in your research, please cite:

Paper Link

@inproceedings{haghbeigi2025adaptive,
  title={Adaptive Potential Fields for Multi-Robot Path Planning with Convergence Avoidance},
  author={Haghbeigi, Morteza and Ordys, Andrzej},
  booktitle={2025 European Conference on Mobile Robots (ECMR)},
  pages={1--6},
  year={2025},
  organization={IEEE},
  doi={10.1109/ECMR65884.2025.11163011}}
}