智能无人售货柜系统设计与实现

Design and implementation of intelligent unmanned vending container system

针对弹簧式自动售货机制造成本高、售卖品类少、用户体验差等问题,文中设计一种智能无人售货柜系统。该系统融合了图像传感器和称重传感器,可实现计重销售与计件销售两种模式。首先,为减少订单处理时间,对购物过程中传感器采集的数据进行预处理,将有效信息传输至服务器;然后,在服务器构建商品识别模型,并设计基于信息熵的多传感器融合算法,完成用户购物过程中挑选商品的自动识别;最后,借助信用支付功能完成自动扣费,实现“扫码开门,自主选购,无感结算”的全新购物模式。通过25.181万笔真实运营订单测试得出,文中设计的系统商品识别准确率为97.76%。对比弹簧式自动售货机,文中系统具有购物流程简单、销售商品种类丰富、制造成本低等优势,更符合商业运营要求,可为无人零售行业提供全新的思路和技术参考。

In allusion to the problems of spring⁃loaded vending machines,such as high manufacturing cost,few sale categories and poor user experience,an intelligent unmanned vending container system is designed.This system contains the image sensor and weighing sensor,which can realize two modes of weight sales and piece sales.The data collected by the sensors during the shopping process is preprocessed,and the effective information is transmitted to the server to reduce the order processing time;the product recognition model is constructed on the server,and the multi⁃sensor fusion algorithm based on information entropy is designed to complete the automatic recognition of the product selected by users during the shopping process;with the help of credit payment function,the automatic fee deduction is completed to realize the new shopping mode of"scan code door⁃opening,self⁃purchase and automatic settlement".After the testing of 251810 real operating orders,the results show that the product identification accuracy of the system designed in this paper is 97.76%.In comparison with the spring⁃loaded vending machine,the system has the advantages of simple shopping process,rich types of sales goods and low manufacturing cost.It is more in line with the requirements of commercial operation,and can provide a new idea and technical reference for the unmanned retail industry.