Title:

Towards Optimal Information Exchange Instants in Multi-Agent Systems

Abstract:

We let agents learn how often to exchange information with neighbors in cooperative Multi-Agent Systems (MASs) such that their user-defined cost functions are minimized. The investigated cost functions capture trade-offs between the MAS local control performance and energy consumption of each agent in the presence of exogenous disturbances. Agent energy consumption is critical for prolonging the MAS mission and is comprised of both control (e.g., acceleration, velocity, etc.) and communication efforts. The proposed methodology starts off by computing upper bounds on broadcasting intervals that provably stabilize the MAS. Subsequently, we utilize these upper bounds as optimization constraints and employ an online learning algorithm based on Least Square Policy Iteration (LSPI) to minimize the cost function for each agent. Consequently, the obtained broadcasting intervals adapt to the most recent information (e.g., delayed and noisy agents’ inputs and/or outputs) received from neighbors and provably stabilize the MAS. Chebyshev polynomials are utilized as the approximator in the LSPI while Kalman Filtering (KF) handles sampled, corrupted and delayed data. The proposed methodology is exemplified in a consensus control problem with general linear agent dynamics.

Bio:

Domagoj Tolic is a postdoctoral researcher at the Department of Electrical and Computer Engineering, University of Dubrovnik. He graduated from the Faculty of Electrical Engineering and Computing (FER), University of Zagreb, with a Master degree (Dipl.-Ing.) in Electrical Engineering (2007), majoring in Control Systems. In addition, he graduated from the Mathematics Department, University of Zagreb, with a Bachelor Degree in Mathematics (2008). Subsequently, he enrolled in the Ph.D. program, the Control Systems major, at the Department of Electrical and Computer Engineering, University of New Mexico (UNM), Albuquerque, NM, as a member of the MARHES (Multi-Agent, Robotics, Hybrid, and embedded Systems Laboratory) research group under supervision of Prof. Rafael Fierro. Upon completing the Ph.D. program (2012), he was a postdoctoral researcher in the Research Centre for Advanced Cooperative Systems (ACROSS) at FER until September 2014. During his time in ACROSS, Dr. Tolic was a visiting researcher under supervision of Prof. Sandra Hirche at the Institute for Information-oriented Control, Technical University of Munchen, from October 2013 until June 2014. His research focuses on stability and estimation under intermittent information for nonlinear and multi-agent control systems. The developed theory is applied to problems in the area of multi-agent robotics.