计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (12): 234-240.DOI: 10.3778/j.issn.1002-8331.1601-0240

• 工程与应用 • 上一篇    下一篇

四阶动力学建模和随机最优控制的UAV目标跟踪

蔡鹏飞1,张正本1,孙  挺2   

  1. 1.河南工学院 计算机科学技术系,河南 新乡 453002
    2.西北大学 可视化研究所,西安 710069
  • 出版日期:2017-06-15 发布日期:2017-07-04

UAV tracking based on fourth order dynamic modeling and stochastic optimal control

CAI Pengfei1, ZHANG Zhengben1, SUN Ting2   

  1. 1.Department of Computer Science and Technology, Henan Institute of Technology, Xinxiang, Henan 453002, China
    2.Institute of Visualization, Northwestern University, Xi’an 710069, China
  • Online:2017-06-15 Published:2017-07-04

摘要: 现存大多数固定翼无人机(UAV)跟踪地面目标的方法需要一个或多个严格的假设,限制了实际应用,针对此问题,提出一种基于四阶动力学建模和随机最优控制的UAV跟踪方法。使用四阶动力学方程对UAV随机动态和目标进行初始化建模;通过一个适当的状态转换概率函数帮助UAV为每一个滚转动作画出蒙特卡罗样本;通过随机优化控制问题的解决确定最优控制反馈策略。仿真实验结果验证了该方法在实际应用中的有效性,与其他优秀方法相比,提出的方法的UAV仰角没有超出瞬时视场角(-152°~32°)的范围,目标没有逃离UAV的视线之外,而其他方法会使目标逃离UAV视线一次或两次。另外,能容忍的最高平稳风速最高达6 m/s,高于其他方法。

关键词: 固定翼无人机, 随机动态, 随机优化控制, 门特卡罗样本, 瞬时视场角

Abstract: Most existing methods of ground target tracking with fixed-wing Unmanned Aerial Vehicle (UAV) require one or more restrictive assumptions, which limit the practical application, for this issue, a tracking method using a UAV based on fourth order kinetics modeling and stochastic optimal control is proposed. Firstly, fourth order kinetic equation is used to model UAV stochastic dynamic and target initialization. Then, an appropriate state transition probability function helps UAV to draw Monte Carlo sample for each rolling movement. Finally, the stochastic optimal control feedback is determined by solving stochastic optimal control problem. Simulation results verify this approach effectiveness in practical applications in comparison with other excellent methods, UAV elevation of proposed method has never exceeded the range of view angle field (-152° to 32°), and the target never escapes out sight of UAV, and other methods can make targets flee sight of UAV once or twice. In addition, the proposed method can tolerate a maximum steady wind at the speed of 6 m/s, which is higher than that of other methods.

Key words: fixed-wing unmanned aerial vehicle, stochastic dynamic, stochastic optimal control, Monte Carlo sample, field of view