Hierarchical Multi-rate Measurement Fusion for Estimation of Dynamical Systems

Fusion algorithms for optimal estimation of dynamical systems are presented. The two algorithms presented, focus on fusion in the presence of asynchronous measurements that are obtained at differing sampling instances and rates. First of these Kalman filter based schemes is based on multiple measurement model switching. The second scheme essentially propagates all the measurements until the update time. The decentralization of the filtering scheme, in which each measurement subsystem is equipped with a dedicated simple estimator and the outputs are combined by a central filter is a characteristic feature of the schemes presented. Features of the algorithms presented are discussed with application on a two wheel differentially driven robot equipped with redundant sensors in a decentralized framework.