Multi-sensor track fusion via Multiple-Model Adaptive Filter

Abstract

A Multiple-Model Adaptive Filter (MMAF) is developed for use in multi-sensor track fusion systems for target tracking. The architecture of hierarchical fusion consists of several local processors and a global processor. Each local processor collects measurement data from a sensor and then using Kalman filter performs tracking function. The global processor utilizes the MMAF which consists of Information Matrix Filter (IMF) with two levels of common process noise and a decision logic switch to aggregate the outputs of local processors. The switch is designed by adopting the modified probabilistic neural network to compute the probability of each IMF for providing the switching capability to respond target dynamics. The resulting filter has better tracking performance than each individual IMF. Simulation results are included to demonstrate the effectiveness of proposed algorithm.

On-Line Bayesian Classifier Design for Measurement Fusion

“…In tracking applications, it is known that the target dynamics may vary rapidly. In order to keep tracking in high accuracy, adaptive state estimation [3,4,5,6,7] are frequently used. In the literature, radial basis function networks [8] have received much attention recently because they provide accurate generalization on a wide range of applications.…”

Decoupled Adaptive Tracking Algorithm for Multi-Sensor Measurement Fusion

“…In real tracking situations, it is known that the target dynamics may vary rapidly. In order to keep tracking in high accuracy, adaptive state estimation [5,6,7,8,9,10] is frequently used. In this paper, as shown in Fig.…”