东南大学学报（自然科学版）2012，Vol.42Issue(z1)：60-65,6.DOI:10.3969/j.issn.1001-0505.2012.S1.013
基于分布式纳什Q学习的多传感器协同目标跟踪
Target tracking by multi-sensor cooperation method based on distributed Nash Q-learning
摘要
Abstract
In order to solve the problem of excessive dependence on the environment model which exists in traditional target tracking algorithm, a multi-sensor cooperation target tracking method based on distributed Nash Q-learning is proposed. The reinforcement learning and distributed Nash Q-Learning theories were analyzed. The multi-sensor cooperative tracking situation was described; the nonlinear models of discrete system were established and its traditional solutions extended Kalman filtering was given. Sensor action and reward function were defined, which are crucial to the learning performance. Reward function was obtained by calculating the trace of prediction error covariance matrix. The probability statistics method based on Bayesian inference was presented to update the Q function. Simulation results of the bearing-only measurements target tracking show that this algorithm can enhance the sensors' environmental adaptiveness, realize the tracking effectiveness and improve the tracking accuracy compared with the traditional filtering algorithm.关键词
目标跟踪/非线性滤波/强化学习/纳什Q学习/分布式控制/多传感器协同/算法Key words
target tracking/ nonlinear filtering/ reinforcement learning/ Nash Q-learning/ distributed control/ multi-sensor cooperation/ algorithm分类
信息技术与安全科学引用本文复制引用
蔡佳,黄长强,高翔,胡杰..基于分布式纳什Q学习的多传感器协同目标跟踪[J].东南大学学报（自然科学版）,2012,42(z1):60-65,6.基金项目
航空科学基金资助项目(20105196016). （20105196016）
