Advanced Algorithms for Prediction, Display, and Visualization of Moving Targets

Toyon Research Corporation proposes to develop an algorithm for the display and visualization of ground vehicle movement predictions based upon sequential Monte Carlo sampling techniques, or particle filters. Ground vehicle movement comprises a vast number of nonlinear elements imposed by operational factors such as terrain, weather, movement characteristics, and tactics that make the prediction problem difficult to formulate. Often such nonlinearities are discarded, but doing so limits the precision of the resulting prediction; an undesired consequence when human operators must make decisions based upon the predictions they receive. Thus, a prediction algorithm is required that provides both a coherent display of the probable movement, and incorporates nonlinear factors that affect the vehicle's motion. Particle filtering methods provide a fitting solution to such an application due to their acceptance of nonlinearities in the dynamic model of the system, and their natural ability to formulate an intuitive visual representation of the uncertainty in the prediction. At the conclusion of Phase I, Toyon will demonstrate a ground vehicle movement prediction algorithm that fuses battlespace information into a particle filter state estimator to provide accurate predicted PDFs for display.