Multi-Robot Task Allocation and Coordination in Dynamic Environments

The work aims to advance the theoretical foundations and practical implementation of  Multi-Robot Task Allocation (information-sharingMRTA) and fleet coordination for dynamic environments. This involves tackling issues like estimating task execution costs, modeling environmental dynamics, selecting task assignment algorithms, and investigating existing planning and representation solutions. A key goal is developing a multilayer architecture that integrates robust task allocation, fleet management, and operations capabilities tailored for volatile, dynamic modelling. 
Ultimately, the research will enable long-term autonomous decision-making and safe learning for large robot teams operating alongside humans across many application domains like logistics, services, exploration, and more.

Research Directions:

• Develop dynamics-aware tools for MRTA and fleet coordination. 
• Identify key challenges like task cost estimation and environmental dynamics modelling. 
• Investigate existing MRTA/coordination solutions and environmental representations. 
• Create a multilayer system architecture for dynamic task allocation and fleet management. 
• Devise methodologies for robust, valuable information sharing within robot teams. 
• Establish online methods to update environment maps.
• Implement hardware-in-the-loop simulations in the robotics lab.

Keywords:

• Multi-Robot Task Allocation Fleet Coordination Reinforcement Learning

Collaboration with FK Group

We are collaborating with the FK Group, known as "Adaptation and Coordination in Dynamic Environments," to leverage Meta Reinforcement Learning (Meta RL). This collaboration focuses on training intelligent agents capable of task allocation in dynamic environments. Meta RL provides a robust framework for developing adaptive and resilient multi-robot systems that can efficiently handle the complexities of dynamic and unpredictable environments.

People Involved:

• Maryam Kazemi Eskeri
• Tomasz Piotr Kucner