Data Fusion & Tracking

Toyon’s Data Fusion and Tracking work varies broadly from the development of fundamental data association and state estimation techniques, to the design of novel approaches for multi-INT data fusion, to the production of application-specific tracking systems for military and commercial use.

Our continuously growing library of algorithms includes a wide-range of traditional techniques for data fusion as well as a collection of proprietary algorithms that expand upon the traditional technologies.

These algorithms include:

• Multi-target tracking algorithms, such as Multiple Hypotheses Trackers (MHT), Joint Probabilistic Data Association (JPDA)-based trackers, and Global Nearest Neighbor (GNN)-based trackers;
• State estimation and filtering algorithms, such as robust and efficient Interacting Multiple Model (IMM) estimators and Particle Filters (PF);
• Feature-Aided Tracking algorithms for exploitation of kinematic and signature information from sensor data; and
• Higher-Level Fusion tools, such as Dynamic Bayesian Networks.

To complement our algorithms research, Toyon owns and operates a small fleet of UAVs to perform hardware-in-the-loop testing and demonstration of our solutions. We also offer this capability as a service for researchers interested in testing their own solutions on a real UAV system.

Exploiting our vast library of algorithms and associated hardware, Toyon continues to address a range of difficult ISR problems with innovative data fusion and tracking solutions. For example, we are currently researching and developing Video-Based Target Tracking algorithms for robust and real-time operation on both fixed-position and small-UAV sensor platforms. This challenging problem demands a comprehensive solution addressing all stages of the video signal processing system including frame registration, target detection, data association and state estimation.