Reza Olfati Saber
Postdoctoral Scholar
Control and Dynamical Systems
California Institute of Technology
 

Title:
 
Distributed Control of Many-Agent Systems: Flocking and Self-Assembly
 
Abstract:
 
Many-Agent Systems are interconnections of large number of physical agents that might or might not communicate with each other.
Some of the examples of such agents include dynamic systems, unmanned autonomous vehicles (UAVs), robots, sensors & actuators,
small parts, and atoms. Many-Agent Systems appear in broad scientific, commercial, and military applications such as distributed atmospheric
sensing, mobile ad-hoc networks, search and rescue operations, reconnaissance and surveillance operations, power networks,
nano-scale self-assembly, and Molecular Dynamics (MD) simulations.  The examples of biological many-agent systems include flocks of
birds, swarms of bees, and biomolecular systems (i.e. proteins).      

 
This talk is divided into three parts. The first part provides and overview of my research on five basic problems that frequently arise in
study of distributed dynamic systems. In the second part, a theoretical framework for flocking/swarming will be introduced.  My future
directions of research are presented in the last part of the talk.  

 
The main body of the talk is devoted to stating the basic elements of a theoretical framework for design and analysis of collective
motion of large number of communicating dynamic agents. This can be viewed as a form of stability and control theory for distributed dynamic
systems under the constraints of computational scalability and low communication costs.  We introduce and analyze a novel flocking algorithm
that extends and unifies three famous heuristic rules of Reynolds in a single equation. The relationship between small-scale self-assembly and
flocking will be also discussed.  Simulation results will be presented that demonstrate performing split/rejoin and squeezing maneuvers
during low-altitude flight of UAVs for hundreds of agents.