Learning Structure, Reusability and Real-time Modeling in Teams of Autonomous Robots

Impact:

• Broad applicability of research on a rich variety of large-scale multirobot systems

• Effective sensing, action, and behavior of robot teams in adversarial tasks

• Real-time response and rapid adaptability in dynamic and uncertain environments

New Ideas:

• Sensor resetting localization for poorly modeled robot movement

• Distributed sensing through communication

• Multi-fidelity behaviors for robust control in the face of variable state accuracy

• Automated recognition of robot behaviors

• Multiagent, multistate learning

• Abstraction learning for layered control

Year 1 Year 2 Year 3

X Development of robot platform

X Software integration

X Motion control and localization

5-10 robot system

Multiagent learning

and control

ᡂ robot system

Multiagent behavior modeling

Vision and communication

X - accomplished

Back to first slide

View graphic version