足球过人战术在采摘机器人路径规划中的应用

The Application of Football Dribbling Pass Tactics in Path Planning for Picking Robot

随着人口老龄化的到来及农村青壮年的流失,实现农业生产自动化的需求越发迫切。采摘具有劳动强度大、技能要求高的特点,是发展自动化的核心领域。为此,以测距传感器作为检测元件,设计了采摘机器人的行走机构及路径规划系统。路径规划系统采用4层模糊神经网络架构,采用足球过人战术建立避障规则和目标导引规则。采用梯度下降法对神经网络进行训练,通过调整连接权系数ωli、高斯隶属函数中心值cjk及宽度值σjk,使实际输出向输出期望快速逼近,当满足阀值要求时,输出机器人速度V和转角Δa。仿真实验结果表明:系统具有良好的路径规划精度,且结构简单、生产成本低,适于大范围推广。

It is urgent to achieve the automation of agricultural production,with arriving of aging population and losing of young adults in rural area.apple picking is the key in automation of agricultural production,because of great labor intensity and demand for excellent skill.Walking mechanism and path planning system was designed,and range sensor was used as detecting element.Path planning system was constituted with 4 levels fuzzy neural network,avoidance and target guiding rule were designed by football dribbling pass tactics.Gradient descent algorithm was used to train fuzzy neural network,according to adjusting Connection weight coefficientωli,and central value cjk,,widthσjk,of gaussian-type membership function,actual output approached to expectation.Velocity V and angular displacementΔθof robot were output,when threshold value was satisfied.Simulation results showed this system had high detection precision for Path Planning,with characteristics of low cost and simple structure.This system could satisfy demand of extension.