This repository implements a distributed multi-UAV coverage path planning system for networked operations based on our paper:

"Decentralized Communication and Energy-Aware Multi-UAV Coverage Path Planning"
Felix Muller and Nick Truttmann
ETH Zurich, 2025

This project is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

This project addresses the challenge of deploying multiple UAVs to efficiently cover an area while maintaining network connectivity and minimizing energy consumption. We implement a novel approach combining:

1. Distributed centroid Voronoi partitioning: Divides the workspace evenly among UAVs while accounting for their relative capabilities
2. Spanning Tree Coverage (STC): Generates efficient coverage paths within each partition
3. Communication Radius estimation: Uses GHS algorithm to create a minimum spanning tree (MST) to estimate the required communication range.
4. Distributed Bayesian Optimization via Thompson Sampling: Optimizes starting positions and seeds to minimize both energy consumption and communication radius

The system is implemented as a distributed multi-agent ROS2 application where each UAV operates independently while maintaining network connectivity.

Our approach achieves:

• Balanced area assignment among homogenous UAVs
• Complete coverage with minimal path overlap
• Optimized energy efficiency
• Scalability to multiple UAVs

Launch the system with four UAVs:

ros2 run controller controller

Monitoring and cleanup:

ros2 node list

killall -9 uav_agent

Each agent uses multiple mixins:

• PartitionMixin: Implements distributed power diagram partitioning
• STCMixin: Implements Spanning Tree Coverage algorithm
• RadiusMixin: Handles communication radius calculations
• EnergyMixin: Computes energy consumption for paths