智能轨迹控制割草机器人设计——基于FPGA神经网络
发布时间：2018-10-23 12:57
【摘要】：为了提高割草机器人自主导航和定位的精确性和智能性,设计了一种新型的基于FPGA神经网络算法的割草机器人。该设计采用FPGA可重构技术,以3层误差反向传播神经网络作为典型的模型来展开;利用成熟的BP算法公式,设计了割草机器人智能控制的模型;利用FPGA技术,设计了割草机器人的硬件系统;最后采用文本输入的设计方法,利用田间试验的方式,对机器人的轨迹规划能力和控制精度进行了验证。试验结果表明:利用FPGA和神经网络模型可以有效地穿越5个障碍物,并可得到满意的轨迹规划结果。将普通的PID控制器和神经网络PID控制器得到的控制结果误差进行了对比,结果表明:神经网络PID控制器得到的割草机器人控制误差明显比传统的PID控制器误差小。该方法为神经网络的硬件实现提供了可靠的理论基础。
[Abstract]:In order to improve the accuracy and intelligence of autonomous navigation and localization of grass mowing robot, a new kind of grass mowing robot based on FPGA neural network algorithm is designed. The design adopts FPGA reconfigurable technology, takes 3-layer error back propagation neural network as the typical model, uses mature BP algorithm formula, designs the intelligent control model of mowing robot, and makes use of FPGA technology. The hardware system of the mowing robot is designed, and the trajectory planning ability and control precision of the robot are verified by using the design method of text input and the way of field experiment. The experimental results show that the FPGA and neural network models can effectively cross five obstacles and obtain satisfactory trajectory planning results. The error of the conventional PID controller and the neural network PID controller is compared. The results show that the error of the neural network PID controller is obviously smaller than that of the traditional PID controller. This method provides a reliable theoretical basis for the hardware implementation of neural network.
【作者单位】： 河南工业职业技术学院;
【基金】：河南省自然科学基金项目(2015ZCB115) 南阳市科技攻关项目(2012GG029)
【分类号】：S817.111;TP242
本文编号：2289312
[Abstract]:In order to improve the accuracy and intelligence of autonomous navigation and localization of grass mowing robot, a new kind of grass mowing robot based on FPGA neural network algorithm is designed. The design adopts FPGA reconfigurable technology, takes 3-layer error back propagation neural network as the typical model, uses mature BP algorithm formula, designs the intelligent control model of mowing robot, and makes use of FPGA technology. The hardware system of the mowing robot is designed, and the trajectory planning ability and control precision of the robot are verified by using the design method of text input and the way of field experiment. The experimental results show that the FPGA and neural network models can effectively cross five obstacles and obtain satisfactory trajectory planning results. The error of the conventional PID controller and the neural network PID controller is compared. The results show that the error of the neural network PID controller is obviously smaller than that of the traditional PID controller. This method provides a reliable theoretical basis for the hardware implementation of neural network.
【作者单位】： 河南工业职业技术学院;
【基金】：河南省自然科学基金项目(2015ZCB115) 南阳市科技攻关项目(2012GG029)
【分类号】：S817.111;TP242
【相似文献】
相关期刊论文 前2条
1 胡建平;严宵月;何俊艺;关静;;钵苗移栽机器人控制系统设计研究[J];农机化研究;2014年12期
2 ;[J];;年期
相关会议论文 前1条
1 康惠春;王建军;马春庆;王卫星;贾天义;闫立峰;尹德秀;范巧变;谷书伟;杨文娟;杨大芳;;大型拖拉机底盘机器人自动喷涂集成系统[A];第十五届全国涂料与涂装技术信息交流会暨商用车、工程机械、轨道交通涂装技术研讨会论文集[C];2012年
相关硕士学位论文 前2条
1 董硕;林区巡防监测机器人研究[D];中国林业科学研究院;2013年
2 崔军;喷雾机器人运动控制与重喷漏喷显示[D];江苏大学;2006年
，本文编号：2289312
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2289312.html
