Multisource-Multisensor Data Fusion

A Stochastic Neuro-Fuzzy System for Multiple Maneuvering Target Tracking

Aptronix, Inc has developed an advanced data fusion technology for target tracking in dense clutter environment. This Automated Multiple Maneuvering Target Tracking (AMMTT) system consists of four subsystems: track initiation, data association, track maintenance and track termination. It is implemented in a hierarchical architecture containing two adaptive filters in parallel and two separate neural networks. To achieve high accuracy in track maintenance, the state fusion approach is proposed that uses all state information and the stochastic neural-fuzzy methodology. To achieve fast data association, a stochastic neural network is developed to compute the joint association probabilities from measurement data. Major advantages of the AMMTT system include (1) rapid response to a wide range of maneuvers; (2) high-precision estimation of target kinematic state; (3) very fast data association; and (4) only two state models are used to track all maneuvers. Computer simulations with fighter jets are illustrated to demonstrate AMMTT’s superior performance and its potential application to general data fusion and tracking problems.

GPS-based Positioning and Navigation Technology

• Fast, accurate positioning
• Multiple track model (MTM)
• No recovery time needed
• No inertia effect of a Kalman filter-based positioning system
• Simultaneous sensor and model error corrections
• Built-in self-adaptation
• Robust performance even when GPS is lost in cities
• Applicable to military (air defense) & civil systems (GPS, air traffic control)

Back to Home, Copyright © 1997. Aptronix, Inc.