Investigator: Stephan Shatara

Collaborator: Dr. John S. Baras (University of Maryland)

Select publications:

• X. Tan, "Decentralized Coordination of Unmanned Vehicles Using Parallel Gibbs Sampling," IEEE Transactions on Automatic Control (under review, 2007)
• W. Xi, X. Tan, J. S. Baras, "Gibbs Sampler-based Coordination of Autonomous Swarms," Automatica (2006)

Collaborative control of multi-agent systems presents itself in different contexts: MEMS sensor and actuator networks, robotic swarms, and automated highway systems. The large scale of such systems calls for a distributed/decentralized approach. Inspired by the emergent behaviors demonstrated by bacteria, insects, and animals, we are exploring control methods that lead to desired, global goals under simple local interactions. One approach we have proposed is based on the theory of Markov Random Fields, and is applied to control of autonomous swarms. We are interested in both the development of efficient algorithms and the analysis of algorithm behaviors. Click here for an animation of self-organization of vehicles into a linear formation.

An experimental testbed, consisting of a group of communicating robotic fish,  is also being developed to study and verify distributed control methods for autonomous swarms. Each robotic fish will be propelled by an electroactive polymer actuator, and be equipped with GPS, Zigbee, microcontroller, and onboard sensors for navigation, control, communication, and sensing. This testbed will allow us to examine high-level motion planning schemes, low-level motion control methods, communication protocols, sensor fusion algorithms, as well as their interactions.