A Deep Learning Maneuvering Target Tracking Algorithm with Ordinary Differential Equations Trajectory Generation

In the field of maneuvering target tracking, most tracking algorithm rely on predefined mathematical models which require a large amount of maneuvering target trajectory datasets. However, due to the lack of sufficient trajectory data before tracking, it is challenging to accurately model the highly variable and uncertain maneuvering motion. This paper proposes a Deep Learning Maneuvering Target Tracking Algorithm with Ordinary Differential Equations Trajectory Generation algorithm (DeepMTT-ODEs), which is based on Long Short-Term Memory (LSTM) networks, to track maneuvering targets. The ODEs are used to construct an offline large-scale trajectory database, and then the DeepMTT algorithm is employed for target tracking. This algorithm effectively addresses the issue of insufficient data in maneuvering target tracking problems. Experimental results demonstrate that the proposed algorithm offers superior tracking performance, faster processing time, and greater adaptability compared to traditional tracking algorithm.