基于模糊神经网络的机器人感知系统多源信息融合的研究  

Research on Information Fusion of Multiple Sources Sensor of Robot Perception System Based on Fuzzy Neural Network

作　　者：杨杰[1] 张明路[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3期41-43,共3页Journal of Hebei University of Technology

基　　金：国家863计划项目资助(2006AA04Z221);河北省自然科学基金项目资助(E2006000030)

摘　　要：提出了利用基于模糊神经网络的传声器阵列和双目摄像机信息融合,实现对声源的精确定位和实时跟踪的方法.该方法建立了五层模糊神经网络,设计了各层的输入和输出,利用BP算法调整模糊神经网络的权重和参数获得不同的隶属函数,从而生成相应的模糊规则实现模糊决策.利用Matlab仿真了机器人的感知过程,得到了实际输出和实验输出的误差曲线,验证了基于模糊神经网络多源传感融合算法对于机器人控制的有效性.A new information fusion decision method about fusing microphone signal and CCD camera signal based on fuzzy neural network was porposed in this paper, which is used for sound source precise location and real-time follow-up. Five levels fuzzy neural network were built and its input and output were designed, the weighing and parameters were re- vised in order to get different membership function, then corresponding fuzzy rule and could execute fusion decision could be got in the new method. The simulation perception processing of robot was completed using Matlab, the error curve between actual output and experiments output was got. The simulation results show that the validity of robot control meth- od based on fuzzy neural network multiple sources sensor information fusion

关 键 词：模糊神经网络 多源传感信息 融合决策 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]