Online Planning with Digital Twins for Heterogeneous Robot Fleets in Mixed Environments with Uncertainty

• Grant number: 24/15649-7
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: December 01, 2024
• End date: November 30, 2027
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Thiago Boaventura Cunha
• Grantee: Matheus Aparecido do Carmo Alves
• Host Institution: Escola de Engenharia de São Carlos (EESC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 18/15472-9 - Hydraulic actuation impedance control for robots with legs and arms, AP.JP

Abstract

Research on mobile robots focused on solving problems in environments where humans and robots coexist has attracted significant public and private investments in recent years. In this context, deploying heterogeneous fleets of intelligent and autonomous robots has demonstrated clear advantages, particularly when performing tasks in hazardous or hard-to-reach areas for humans. These systems' high adaptability allows the robots to organize themselves effectively to solve complex tasks while providing support to the humans present in the environment. However, navigation and action planning in these contexts require the robots to be capable of quickly and efficiently adapting their behaviour to handle uncertainties, reduce operational costs, and ensure the safe and effective execution of their missions, avoiding damage to themselves, the environment, and the people involved. Therefore, this research project proposes the development of a robust, decentralized, real-time planning algorithm that enables different types of robots to collaborate toward a common goal while supporting distributed tasks in mixed environments. To ensure the feasibility of this application in real-world scenarios, an embedded, lightweight, and optimized solution will be created, combining reinforcement learning techniques with Digital Twin models. Recognizing the potential benefits of this research, Petrobras supports this project and plans to test and validate the proposed solution on its oil platforms. Overall, this research aims to address gaps in the existing literature and foster advancements in robotics research in Brazil to tackle and resolve concrete challenges faced by the national industry.