Analyzing Swarms a Stochastic Systems Approach to Studying Swarm Behavior

Speaker:
Kristina Lerman
Information Science Institute (ISI)
University of Southern CaliforniaAbstract:Swarms are decentralized systems composed of many simple agents (e.g., robots) with no central controller. Instead, swarms are self-organizing: constructive collective behavior emerges from local interactions among the agents and between agents and their environment. Though the behavior of an individual agent in a swarm can be considered to be stochastic and unpredictable, the collective behavior of many such agents has a simple probabilistic description. We present a formal methodology for studying the behavior of swarms. We derive a class of mathematical models that describe the dynamics of swarms and show how the models can be written down by examining details of the individual agent behavior. This methodology is valid for certain types of agents that can be modeled as Markov processes (of varying complexity). Though at first glance this may appear overly restrictive, most of the currently studied or proposed swarm systems are based on Markovian agents. Like all formal approaches, the stochastic methodology has limitations, but as applications of the approach to the robotics domain illustrate, it can also be quite useful in predicting collective behavior. Bio:
Kristina Lerman is a research assistant professor in Computer Science at the University of Southern California and a project leader at the USC Information Sciences Institute. She received a Ph.D. in physics from UC Santa Barbara in 1995, where she studied complex behavior of spatially extended non-equilibrium systems. She has received funding from NSF and DARPA to apply physics-based mathematical methods to the study of multi-agent systems, such as distributed robot systems.

Location: Kaprielian Hall (KAP) - 203

Audiences: Everyone Is Invited

Contact: Evangeline Reyes