Collaborative Research: CISE-MSI: DP: CNS: An Edge-Based Approach to Robust Multi-Robot Systems in Dynamic Environments

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2). Multi-robot systems consist of autonomous robots interacting in a shared environment to achieve common goals. They are widely used in real-world application domains such as transportation, disaster management, as well as warehousing and manufacturing. This project develops an efficient, robust, and secure multi-robot system, called EdgeRobot. EdgeRobot establishes an edge computing based architecture and algorithmic framework to facilitate multi-robot collaboration and coordination in dynamic environments. This work provides new model, architecture, and theory for coordinated multi-robot systems. In addition, this project builds research capacity, sustainable for training underrepresented students via the partnership of six geographically diverse minority-serving institutions in the United States: the University of Houston-Clear Lake (South), the University of Michigan Flint (North), CUNY-New York City College of Technology (Northeast), Morgan State University (East), San Francisco State University (West), and California State University Dominguez Hills (West). The cross-institutional collaboration not only boosts research capacity in all six participating institutions but also provides integrative research and education experience to their underrepresented minority students. Ultimately, this project establishes and exemplifies an effective collaboration model for training and educating underrepresented students from geographically diverse minority-serving institutions. This project consists of the following three research thrusts. First, the novel edge computing infrastructure provides optimal and location-aware computing services for collaborative robots to achieve their common goals. Besides, reinforcement learning-based algorithms solve the multi-robot scheduling and routing problems, modeled as variants of the prize-collecting traveling salesman problem. Second, in tasks requiring collaborative actions, such as cooperative target tracking, multi-agent reinforcement learning enables teams of robots to operate, learn, and adapt in dynamic and human-populated environments robustly and safely. Third, integrating modern cryptographic and security primitives secures the collaboration among edge nodes in multi-robot systems. Consequently, the interface between EdgeRobot and its human team members builds a shared autonomy model. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.