动态环境下基于实际隶属函数的移动机器人路径规划  

Path Plan for Mobile Robot Under Dynamic Enviroment Based on the Actual Membership Function

作　　者：熊开封[1] 张华[2] 

机构地区：[1]西南科技大学国家级综合性工程训练中心,四川绵阳621010 [2]西南科技大学信息工程学院,四川绵阳621010

出　　处：《科技通报》2015年第9期168-173,共6页Bulletin of Science and Technology

基　　金：四川省科技支撑计划项目(项目编号:2013GZX0152);西南科技大学青年基金项目(项目编号:11zx3103)

摘　　要：为优化模糊神经网络的实时性、学习速度、收敛性、稳定性,在移动机器人局部路径规划中构建了基于实际隶属函数T-S(Takagi-Sugeno)模型的改进型模糊神经网络。对外部环境信息用多传感器(超声波、摄像头)采集并优化,将机器人横纵坐标及行进方向作为输入、机器人下一步行进方向及速度作为输出,以便机器人实现局部路径规划;结合动态环境下机器人路径规划的实际,综合考虑二维直角坐标体系下机器人、障碍物的位置、速度及运动方向等实时信息,推导出一种新的具有实际含义的隶属函数作为避碰隶属函数,并通过对比隐含层节点数对网络相对误差的影响来确定隶属函数层节点数,构建五层T-S型模糊神经网络;在此基础上应用改进型误差反传学习算法,通过matlab模拟实验仿真验证及对比分析,表明了改进型网络在优化网络实时性、学习速度、收敛性、稳定性方面有良好的性能。To optimize the fuzzy neural network real time,learning speed, convergence and stability, mobile robot local path planning improved fuzzy neural network model based on the actual membership function t-s type(Takagi-Sugeno)is presented.This method is to use multiple sensors(ultrasonic, camera) to collect the external environment information and optimize them, the robot coordinate transverse, longitudinal axis coordinates and direction were used as input variables, the robot move to the next step and direction speed were used as the output variable, this helps to robot local path planning implementation.This method is to make the gaussian function deformed as membership function,and by comparing the relative error of the node number of implicit layer of network to determine the membership function layer node number,to construct the five layers of the T-S type fuzzy neural network.On the basis of the improved error back propagation, simulation verification by matlab simulation and comparison analysis, show that the improved fuzzy neural network has a good performance in optimization of network learning speed, convergence, stability.

关 键 词：T-S型模糊神经网络 局部路径规划 实际隶属函数 MATLAB仿真 优化 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]