基于水平集方法的动态目标跟踪路径最优仿真  

Route-optimal Simulation for Dynamic Object Tracking Based on Level-set Method

作　　者：李凤玲[1,2] 申群太[1] 荣见华[2] 

机构地区：[1]中南大学信息科学与工程学院控制理论与工程,长沙410007 [2]长沙理工大学汽车与机械学院,长沙410076

出　　处：《系统仿真学报》2009年第5期1314-1316,1319,共4页Journal of System Simulation

摘　　要：两移动平面运动体因有各自的智能控制器,因此当它们在跟踪控制时,双方必然及时根据对方的策略改变各自的控制策略(方向和速度大小),来达到一种真正的双边最优(主动跟踪者最短时间的追上被跟踪者,而被跟踪者最大化时间避免被追上)。首先将平面上任意运动体简化成带速度和加速度的质点,然后利用追逃微分对策理论建立起此跟踪路径最优问题的哈密顿-雅可比-贝儿曼偏微分数学模型。对这种双边极值的难点问题,提出一种水平集数值求解方法。并利用水平集的偏微分方程与哈密顿-雅可比-贝儿曼偏微分方程结构相似性得到水平集的速度函数。对一平面上两无人驾驶飞机的动态跟踪双方进行仿真,仿真编程简单,解决了双边极值的求解。An effective numerical level-set method was proposed for the solution of two plane-dynamic objects＇ routeoptimal tracking, which have two intelligent controller and can change its controlling decision when they are tracking. Firstly route-optimal problem for dynamic object tracking was translated into two-persons zero-sum pursuit-evasion differential games. Then the time-optimal solution was turned into getting an approximate value function of the Hamilton- Jacobi- Bellman partial differential equation （PDE）. The similar math structure of Hamilton-Jacobi-Bellman PDE and level set PDE were used and the implicit velocity function was got. The simulation results of two plain airplane tracking prove that level set method is effective.

关 键 词：动态目标跟踪 路径最优 水平集 追逃微分对策 

分 类 号：TP297.1[自动化与计算机技术—检测技术与自动化装置] TP301.6[自动化与计算机技术—控制科学与工程]