DMCHBA_Hungarian-Based for Task Allocation of MultAgent Syst

Decentralized multi-task allocation algorithm for multi-agent sys with proven efficiency and conflict-free performance.

This repository provides the official MATLAB implementation of the Distributed Matching-by-Clone Hungarian-Based Algorithm (DMCHBA) — a novel decentralized task allocation strategy designed for multi-agent systems operating in environments with limited communication and a high number of tasks relative to agents. This algorithm was published in the IEEE Transactions on Robotics and selected for oral presentation at the 2024 IEEE International Conference on Robotics and Automation (ICRA) in Japan.

The DMCHBA framework addresses one of the most challenging scenarios in robotics: assigning multiple independent tasks to a smaller number of agents, in a way that ensures conflict-free, efficient, and scalable coordination. Unlike conventional centralized algorithms or distributed approaches with heavy communication overheads, DMCHBA leverages agent cloning, pseudo tasks, and the Hungarian method in a fully distributed and asynchronous communication setting.

DMCHBA is particularly powerful for environments where:

• Agents cannot access global task information from a central node,
• Communication is limited to neighbors in a dynamic or static network,
• Optimal or near-optimal task allocation is required at scale.

This MATLAB release includes full source code, simulation data, performance plots, and two custom task sequencing strategies — the Heuristic Local Path Planning Algorithm (HLPPA) and Naive Local Path Planning Algorithm (NLPPA) — developed to enhance the order in which agents execute their assigned tasks.🔑 Key Features:

• Distributed coordination: Compatible with dynamic or static network topologies using local-only communication
• Cloning-based assignment: Generates square cost matrices by replicating agents virtually (cloning), avoiding recursive Hungarian calls
• Number of Tasks flexibility: Supports any number of tasks, even when not evenly divisible by the number of agents
• Path planning: Optional local TSP-based planning via HLPPA or NLPPA
• Monte Carlo Simulation validation: Outperforms CBBA, CBHA, and DRHBA in both total cost and execution time across 1000 Monte Carlo simulations

📁 What's Included:

• Main implementation of the DMCHBA algorithm (dmchba.m, main.m)
• Utility functions for task-agent cost generation and result visualization
• Local path planning methods
• Input data (.mat files) for simulation runs
• Performance comparison plots and sample result figures
• Complete README, License (MIT), and citation details

📄 Citation:

If you use this code in your research or publications, please cite:

A. Samiei and L. Sun,

“Distributed Matching-by-Clone Hungarian-Based Algorithm for Task Allocation of Multi-Agent Systems,”

IEEE Transactions on Robotics, 2023.

https://doi.org/10.1109/TRO.2023.3335656

This research and implementation were developed by Dr. Arezoo Samiei as part of her Ph.D. dissertation and ongoing academic contributions to decentralized robotics and autonomous systems.

👤 Author:

Arezoo Samiei, Ph.D.

Phone: [phone]

Email: [e-mail] | [e-mail]

LinkedIn: https://www.linkedin.com/in/arezoo-samiei

Google Scholar:

https://scholar.google.com/citations?user=XXXX (← replace XXXX with your real Scholar ID)

ORCID:

https://orcid.org/0000-0001-5812-5701