Trajectory Tracking and Prediction Based on Bi-directional Attention Network
ZHAO Pengyue
LIU Huajun
Abstract：Data driven trajectory tracking and prediction(TTP)methods,such as long short-term memory networks(LSTM),bidirectional LSTM(Bi LSTM),and other deep neural network models,have achieved significant improvements,overcoming the shortcomings of traditional model-based methods in modeling sensor and motion uncertainties.This article proposes a bidirectional attention network(BiAN)for representing long-distance trajectories and maneuvering target movements,as well as learning physi-cal motion prediction from sensor noise measurements.This article establishes a large-scale trajectory tracking and prediction(TTP)dataset that includes pedestrian and vehicle cases as a new TTP benchmark for deep neural network training and validation.The BiAN model proposed in this article has been validated on public datasets(ETH+UCY)and a new benchmark TTP dataset,and the results show that the BiAN model proposed in this article outperforms existing state-of-the-art deep models in RMSE,ADE,FAD and other aspects.
Keywords：target trackingtrajectory trackingtrajectory predictiondeep learning
Publication Date：2025-05-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：9( 1293-1300,1374 )