The objective of this project is to design algorithms that allow teams of simple mobile robots to complete a wide range of tasks reliably. The key insight is that effective planning is possible for such teams, even in spite of significant uncertainty stemming from both sensing and motion. This approach builds upon existing work on minimalism for single robots, but must also overcome substantial complications that arise from coordination and communication between the robots. A distinguishing feature of this work is that the problems are complex and nontrivial at multiple scales: Planning for the multi-robot teams cannot be fully decoupled from the planning and control issues for individual robots.
The project combines three related research endeavors. First, it investigates techniques for representing each robot's uncertainty about its own state, and about the state and knowledge of the other robots. Second, it develops energy-efficient and robust strategies for communication between robots. Third, it applies these techniques in specialized planning algorithms to allow robot teams to complete their tasks in a decentralized manner.
This research will result in a collection of new algorithms that will allow teams of simple robots to manage uncertainty, communication, and planning in order to complete broad classes of tasks. These algorithms will enable simpler, more autonomous teams of robots to be deployed with less expense. Such robots will have significant positive impact on many sectors of our society, including transportation, space exploration, and agriculture. Broader impact will include development and distribution of "covertly educational" game software for middle school students, and training of undergraduate and graduate student researchers.
