Real-time Multisensor Tracking and Correlation of Surface Targets

Toyon Research Corporation proposes to develop a tracking system that is able to accept measurements from radars and EO/IR sensors to track surface targets. Our approach will be to use a Sequential Monte Carlo Technique, sometimes called Particle Filter, to filter measurements and compute association likelihoods. We will use the association likelihoods in an assignment matrix to perform data association using a modified Auction Algorithm. The particle filter will be designed to accept signatures as well as kinematic information, should it be available. We will explore several different options for simulating the motion of objects on the sea surface. This will include the assumption of a flat surface with process noise as well as approaches that seek to estimate the time-varying surface around each target. We will develop a prototype algorithm that we will evaluate and demonstrate using a computer simulation. We will do this in Phase I by making use of existing particle filter software that we have already developed for tracking ground targets. We propose to implement the filter in a Toyon-developed simulation, called SLAMEM, that has models of radars, EO/IR sensors, UAVs, and surface targets.