Hierarchical Multi-Rate Measurement Fusion in Estimation of Dynamical Systems
- Additional Document Info
- View All
Fusion algorithms for optimal estimation of dynamical systems using measurements from multiple sensors are discussed. The algorithms fall under the general classification of measurement fusion and state fusion. We develop a hierarchical framework to handle the fusion problem by grouping sensor sets and arranging them in to a parallel cascade. The cascade outputs are subsequently handled by a central filter. Two alternative options are investigated for state fusion, where the filter bank associated with individual cascade is assumed to give a state estimate. The first option involves the central filter using the state estimates as full state measurements with known uncertainty to derive a subsequent optimal state estimate. The second involves an optimal weighted average approach where the central state best estimate is defined as a weighted average of the individual cascades. A measurement fusion scheme, that is computationally faster is also considered which involves switching between measurement models depending on the availability of measurements. All the algorithms presented are inclusive of the most general case of asynchronous operation of the sensor subsystems (estimators included). The measurement fusion scheme is then applied to the problem of estimation of position and orientation of the planar motions executed by a two wheel differentially driven robot from nonhomogeneous and asynchronous measurements made from redundant sensing systems in a decentralized framework.
author list (cited authors)
Majji, M., Davis, J., & Junkins, J.