Uncertainty Modeling of Target Locations from Multiplatform and Multisensor Data     Sponsor: AFRL Project period: 1999 - 2000 Principal Investigator: Dr. Ron Li Project Team: D. Reed and J. Blevins

Background

Target acquisition and location is one of the most mportant factors to mission success. Sensors involved in a real-time target acquisition system are from multiple platforms and of various classes such as SAR, laser, infrared, electron-optic array, hyperspectral, GPS, INS and others. Data acquired by this kind of integrated sensor system can be characterized by its vast volume, comprehensive spatial relationships, and associated temporal dynamics. To meet the goal of accurate target location a good understanding of uncertainty characteristics of individual sensors and their collective impact on the target location is needed. To solve the problem of multiplatform and multisensor based target location, uncertainty modeling is a key issue. A conceptual design and strategy has been developed to address a series of broad issues on analysis of target location uncertainty, simulation of sensor-target correlation, uncertainty minimization, and optimization of platform and sensor configuration. In this proposed research, efforts will be concentrated on modeling locational uncertainties in multiplatform and multisensor based targeting.

Objectives

• To develop an innovative mathematical model for characterizing individual and collective sensor and sensor system uncertainties;
• To enhance target location accuracy for pre-mission analysis and mission planning;
• To study accurate target location by integrating SAR, optical imaging data, and other data sources;
• To implement the above uncertainty modeling algorithm in a computer software system and to test it using data from AFRL; and
• To analyze target location estimation and uncertainty modeling results for preparation of the development of a dynamic uncertainty modeling system.

We propose to develop an innovative mathematical model for characterizing uncertainties of individual sensors as well as those associated with sensor and estimated target locations. A data set from AFRL including ground control data, images from an optical sensor, and images from a synthetic aperture radar sensor from different platforms will be processed. The model will use all observations available to give an optimal estimation of target locations. Uncertainties involved in sensors and locations estimated from individual and integrated configurations will be investigated to validate the developed model. Such a model will make a significant contribution to the overall strategy of multiplatform and multisensor real-time targeting.