A Highly Maneuvering Target Tracking Algorithm Based on Continuous-Time Target Trajectory Function
- 20 January 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-12
Abstract
To address the challenge of accurately tracking highly maneuvering targets in complex environments, this article proposes a continuous-time target tracking algorithm based on the trajectory function of time (T-FoT). Unlike traditional discrete-time filtering methods that rely on predefined state-space models (SSMs), the proposed approach represents target motion as a continuous function of time and position and employs an incremental radial basis functions neural network (RBFNN) to approximate the trajectory in real time and obtain an explicit T-FoT expression. To enhance trajectory smoothness and stability, a second-order difference regularization term is introduced to constrain local curvature, and a dynamic basis-center adaptation mechanism is designed to improve long-horizon tracking performance. Furthermore, by integrating the recursive least squares (RLSs) method into the RBFNN framework, the prediction uncertainty is explicitly quantified, enabling interpretable and reliable confidence assessment. Simulation results show that the proposed method achieves superior accuracy, smoothness, and robustness under strong maneuvering and noise conditions, and the estimated trajectory consistently remains within the 95% confidence interval (CI), demonstrating strong anti-interference capability and reliable uncertainty modeling.Keywords
Funding Information
- National Key Research and Development Program of China (2024YFC3014400)
- China Scholarship Council
- Civil Aviation Education Talent Program of China (MHJY2025010)
- Civil Aviation Education Talent Program of China (MHJY2024036)
- Funds for Civil Aviation Administration of China (CAAC) the Key Laboratory of Flight Techniques and Flight Safety (FZ2025ZX31)
This publication has 28 references indexed in Scilit:
- On consistency and stability of distributed Kalman filter under mismatched noise covariance and uncertain dynamicsAutomatica, 2023
- From target tracking to targeting track: A data-driven yet analytical approach to joint target detection and trackingSignal Processing, 2022
- Long short-term memory-based deep recurrent neural networks for target trackingInformation Sciences, 2019
- Fitting for smoothing: A methodology for continuous-time target track estimation2016
- Learning GP-BayesFilters via Gaussian process latent variable modelsAutonomous Robots, 2010
- GP-BayesFilters: Bayesian filtering using Gaussian process prediction and observation modelsAutonomous Robots, 2009
- GP-UKF: Unscented kalman filters with Gaussian process prediction and observation models2007
- Target tracking in maneuver centered coordinates2003
- Prediction of noisy chaotic time series using an optimal radial basis function neural networkIEEE Transactions on Neural Networks, 2001
- Use of a kinematic constraint in tracking constant speed, maneuvering targetsIEEE Transactions on Automatic Control, 1993