Evolutionary-based Coordination of Multi-Robot Systems with Dynamic Constraints; Examining the Socio-Economic and Societal Impact of Multi-Robot Systems for Infrastructure Monitoring and Maintenance

Vallarino, Jose, School of Engineering and Applied Science, University of Virginia
Bezzo, Nicola, EN-SIE, University of Virginia
JACQUES, RICHARD, EN-Engineering and Society, University of Virginia

In the increasingly automated landscape of large-scale industrial operations, such as manufacturing facilities and power plants, the deployment of multi-robot systems presents a potent solution to enhance efficiency and safety. The paper "Evolutionary-based Coordination of Multi-Robot Systems with Dynamic Constraints" explores the development of a sophisticated framework for the automated coordination of unmanned ground vehicles (UGVs) tasked with infrastructure inspection and monitoring.
At the heart of the research is the challenge of optimal task allocation and path planning among multiple robots, considered within the framework of the multiple Traveling Salesman Problem (mTSP). We introduce an innovative solution that leverages evolutionary algorithms to dynamically allocate tasks considering factors like priority, energy efficiency, and time constraints. This method not only optimizes the operational efficiency of robot teams but also significantly minimizes human intervention in hazardous environments.
The methodology incorporates a digital twin of the environment, which acts as a pivotal tool for planning and simulation. By using A* and artificial potential field methods, the system achieves optimal path planning and navigation, ensuring that robots can navigate cluttered or complex environments effectively. This setup is complemented by a human-in-the-loop system that allows for adjustments based on real-time changes in the environment or task priority, enhancing the adaptability and responsiveness of the robotic system.
The effectiveness of the proposed solution is demonstrated through simulations using the Gazebo/ROS framework for UGVs. These simulations validate the system's ability to handle realistic, dynamically changing industrial environments efficiently. The research introduces several key innovations:
1. Advanced Task Assignment: By integrating priority, energy, and time considerations, the system efficiently manages the coordination among multiple robotic agents, ensuring that tasks are performed optimally with minimal overlap or redundancy.
2. Use of Digital Twin Technology: The digital twin serves as a critical component for real-time operational planning, allowing robots to adapt to new information and navigate around potential obstacles.
3. Human-in-the-Loop Prioritization: This feature introduces a layer of human oversight that enhances the system’s flexibility, making it robust against unexpected changes in the operational field.
The paper also discusses the implications of these technologies for the future of robotic systems in industrial settings. We suggest that future work could explore the inclusion of more diverse constraints, such as varying capabilities among robots and the integration of human tasks alongside robotic operations. Additionally, real-world applications and further scalability of the system are identified as essential steps toward the practical deployment of these technologies.
This work not only advances the theoretical understanding of multi-robot coordination but also provides a practical framework that can be directly applied to enhance the efficiency and safety of industrial operations. The combination of evolutionary algorithms, digital twin technology, and human-in-the-loop prioritization represents a significant step forward in the field of robotics, offering a scalable solution to complex operational challenges in large-scale industrial settings.

BS (Bachelor of Science)
Multi-Robot Systems

School of Engineering and Applied Science

Bachelor of Science in Systems and Information Engineering

Technical Advisor: Nicola Bezzo

STS Advisor: Richard Jacques

Technical Team Members: Matthew Heeter, Vihar Shah, Patrick Sherman, Lauren Bramblett

All rights reserved (no additional license for public reuse)
Issued Date: