Scholarlink

<p>To address the challenge of trajectory prediction for highly maneuverable aerial targets, this paper introduces the novel Intention-guided Interactive Multiple Model (IGIMM) method. Following a “recognize-then-predict” approach, this model first performs a probabilistic classification of the target’s flight behavior, which facilitates adaptive modeling and enables high-precision prediction for different motion trajectories. This method yields a dramatic performance boost, demonstrated through rigorous testing on simulated and public data: it slashes long-term prediction error by 45% and, within a 25-second window, cuts the root mean square error (RMSE) from 0.53&#xa0;km down to just 0.19&#xa0;km. The model’s robustness is confirmed by extensive Monte Carlo simulations, which consistently show a mean prediction error below 1&#xa0;km. Compared to traditional algorithms like IMM and AGIMM, the IGIMM method effectively solves the sluggish convergence problem when a highly maneuverable target ceases maneuvering. This research provides a significant methodological advancement for aerial tracking and interception, offering powerful theoretical and practical utility.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Interacting multiple model behavior-guided multi-model trajectory prediction for highly maneuverable aerial targets

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月24日

查看全文
  • Panghe Qiu,
  • Nan Zhou,
  • Chunmei Li,
  • Dongjing Cao

摘要

To address the challenge of trajectory prediction for highly maneuverable aerial targets, this paper introduces the novel Intention-guided Interactive Multiple Model (IGIMM) method. Following a “recognize-then-predict” approach, this model first performs a probabilistic classification of the target’s flight behavior, which facilitates adaptive modeling and enables high-precision prediction for different motion trajectories. This method yields a dramatic performance boost, demonstrated through rigorous testing on simulated and public data: it slashes long-term prediction error by 45% and, within a 25-second window, cuts the root mean square error (RMSE) from 0.53 km down to just 0.19 km. The model’s robustness is confirmed by extensive Monte Carlo simulations, which consistently show a mean prediction error below 1 km. Compared to traditional algorithms like IMM and AGIMM, the IGIMM method effectively solves the sluggish convergence problem when a highly maneuverable target ceases maneuvering. This research provides a significant methodological advancement for aerial tracking and interception, offering powerful theoretical and practical utility.