Maneuvering Target Tracking Using Simultaneous Optimization and Feedback Learning Algorithm Based on Elman Neural Network.

Huajun Liu,Liwei Xia,Cailing Wang

doi:10.3390/s19071596

Abstract

Tracking maneuvering targets is a challenging problem for sensors because of the unpredictability of the target’s motion. Unlike classical statistical modeling of target maneuvers, a simultaneous optimization and feedback learning algorithm for maneuvering target tracking based on the Elman neural network (ENN) is proposed in this paper. In the feedback strategy, a scale factor is learnt to adaptively tune the dynamic model’s error covariance matrix, and in the optimization strategy, a corrected component of the state vector is learnt to refine the final state estimation. These two strategies are integrated in an ENN-based unscented Kalman filter (UKF) model called ELM-UKF. This filter can be trained online by the filter residual, innovation and gain matrix of the UKF to simultaneously achieve maneuver feedback and an optimized estimation. Monte Carlo experiments on synthesized radar data showed that our algorithm had better performance on filtering precision compared with most maneuvering target tracking algorithms.

Highlights

Target tracking is a fundamental and critical task in many sensor-based practical applications including radar-based tracking [1], sonar-based tracking [2], wireless sensor networks [3], video surveillance [4], navigation [5], and mobile robotics [6]
From the viewpoint of statistics, maneuvering targets are often modeled as jump Markov linear systems (JMLS) where the maneuver of the target is modeled as a finite-state Markov chain, and its continuously varying state evolves according to an underlying model that switches among a set of operating models controlled by a Markov chain at each sampling instance [7]
In our tracking practice analysis, we find that if the scale of the covariance matrix Qk is smaller, the filtering is more accurate, but when target maneuvering, the scale of Qk should be large enough to ensure that the filter is not be divergent

Summary

Introduction

Target tracking is a fundamental and critical task in many sensor-based practical applications including radar-based tracking [1], sonar-based tracking [2], wireless sensor networks [3], video surveillance [4], navigation [5], and mobile robotics [6]. Classical methods for maneuvering target tracking include two main components: (i) maneuver modeling, the stochastic process assumption for an unpredictable maneuver behavior; and (ii) maneuver compensation, the correction of target state estimates to allow for the maneuver. The first one is based on maneuver dynamics modeling, which is based on the concept of motion-origin uncertainty, and most methods are based on the stochastic process assumption of an unknown acceleration component, such as the Singer model [8], Jerk model [8] and Sensors 2019, 19, 1596; doi:10.3390/s19071596 www.mdpi.com/journal/sensors

Methods

Findings

Discussion

Conclusion