一种基于机器人听觉的多声源跟踪策略  

A Multi-sound Source Tracking Strategy Based on Robot Hearing

作　　者：郝为民[1] 吕晓玲[1] 

机构地区：[1]河北工业大学机械工程学院,天津300130

出　　处：《河北工业大学学报》2009年第5期28-31,共4页Journal of Hebei University of Technology

基　　金：863计划资助项目(2006AA04Z221)

摘　　要：针对多运动声源跟踪问题,提出了一种将波束形成算法和卡尔曼滤波相结合的多运动声源跟踪策略.采用波束形成算法对空间声能量进行搜索计算目标可能的方位,通过不同窗长的并行卡尔曼滤波器预测目标各自在下一时刻的方位,从而获得目标的同一性.仿真与在室内环境下的实际运行结果验证了该方法的有效性.A novel multi-sound source tracking strategy was proposed in this paper. Tracking strategy is composed of two parts. Beam forming algorithm is adopted to calculate the sound wave energy to search the target position, parallel Kalman Filter of various windows is used to forecast the next position of target so to confirm whether it is the same target. Simulation and real experiment results in room environment prove that this tracking strategy is efficient.

关 键 词：机器人听觉 卡尔曼滤波 多声源跟踪 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]