Pequito S., Aguiar A.P., Sinopoli B., Gomes D.A.

International Conference on NETwork Games, Control and Optimization, NetGCooP 2011


This paper introduces Mean Field Games (MFG) as a framework to develop optimal estimators in some sense for a general class of nonlinear systems. We show that under suitable conditions the estimation error converges exponentially fast to zero. Computer simulations are performed to illustrate the method. In particular we provide an example where the proposed estimator converges whereas both extended Kalman filter and particle filter diverge.