For the industry that focuses on unmanned retail, January 21 is definitely a memorable day: on this day, the long-rumored Amazon unattended retail store "Amazon Go" is finally open to the Amazon headquarters in Amazon, meaning Unmanned retailers are welcoming a "big player." On the same day, in Hangzhou, China, Alipay's parent company Ant Financial also launched its unmanned retail experience store WithAnt.
Xiaobian learned that Amazon Go and Ant WithAnt are currently using the "sensor" and "machine vision" mashup technology solutions. However, some industry experts said that once the cost-effective machine vision solution matures, it will become the mainstream of unmanned retail.
Opening of two unmanned retail stores
How is “unmanned retail†defined? According to Amazon's practice results, Amazon Go, the initial answer to this question is: Unmanned retail is not really a staff, but a new type of store that does not require a checkout counter. Users use the Amazon Go app to enter the store, and there is no need to wait long queues in front of the checkout counter after purchasing the item.
In addition to common commodities, cooked foods that were previously considered unsuitable for sale in unattended retail stores are also included in Amazon Go's services. Amazon Go currently offers ready-to-eat breakfast, lunch, dinner and snacks. At the same time, Amazon Go also offers a chef-designed Amazon meal plan, and consumers who choose a package can make a dinner for two in 30 minutes.
On the same day that Amazon Go was officially opened, in Hangzhou, China, Alipay's parent company Ant Financial also launched the first unmanned experience store WithAnt, which mainly sells ant Jinfu peripheral derivative products in the Z space of its new office building.
In the WithAnt unmanned experience store, customers can use Alipay to scan the code and then walk into the store. Want to know the product details, put it on the interactive shelf to see the information on the LED screen on the wall. The selected items can be placed in a box. After the price is automatically recognized by the sensor, the customer can take it and take it through Alipay. According to the official information of ants, the box placed after the selection of goods is a non-inductive payment terminal. Test data shows that it can accurately identify 50 different items in two seconds.
The sensor is still essential
According to Xiaobian, there are basically two major solutions to how to realize the deduction in unmanned retail: sensors (such as RFID chips for goods) and machine vision technology. In terms of the practice of Amazon and Ant Financial, the current technology giant seems to be more willing to adopt the technical solution of sensor and machine vision mashup, and the reason behind it is that machine vision technology is still immature.
For example, the visual visual unmanned retail genre represents Amazon Go's one-and-a-half-year test, and it is only now open because computer vision is always unsatisfactory in the complex environment where multiple people are shopping at the same time. At Amazon Go, when a user is shopping, machine vision will determine whether the user is taking out the goods or putting the goods. When the user misplaces the goods, the Amazon Go system will recognize the misplaced items and the pictures in the database. Staff should be prompted to organize when identifying. Finally, Amazon Go's shelves are equipped with infrared sensors and pressure sensing devices and load sensors that confirm the removal of the product.
In the ant gold clothing WithAnt unmanned experience store less than 30 square meters of space, the same set of more than a dozen cameras, their existence is to identify who took what goods, put back what goods in the middle, when leaving the store How much is the deduction.
Ant Financials admits in a technical paper provided to Xiaobian that machine vision is currently not very reliable. The data shows that in the "eyes" of its computer vision algorithm model, the customer is a pile of lines, and the goods are framed with a similarity of a few percent. If it is a wrinkled notebook, similar The degree will drop, which means that the chance of the machine not recognizing will rise.
Machine vision
Will it be mainstream?
Where will unmanned retailers go in the future? Industry experts believe that machine vision will become the mainstream operation of unmanned retail based on cost and promotion considerations. Of course, the premise is that machine vision is technically mature, otherwise the cost problem is equally difficult to solve.
For example, in Jiang Guofei, vice president of ant Jinfu and head of technical laboratory, computer vision will become the mainstream technology direction for unmanned retail in the future, because sensors will increase operating costs and transform the supply chain, and once computer vision is mature, Can scale up and promote.
In addition, the computer-visually captured details of the goods and people, through the analysis of the algorithm model, can help the store owner to better understand the guests and the store, and make targeted decisions. For example, is it necessary to adjust the position of the explosion item; whether it is to optimize the shelf and merchandise placement strategy according to the route that the user likes most; if the sales performance during the peak period of the passenger flow is flat, is it considered? Replace the recommended items during the peak period; whether the supermarket supply chain needs to be optimized once in a while.
“If you refer to the five-level technical drawing of driverless technology, the fifth stage of unmanned retail technology is pure computer vision technology. No one can come up with a mature technical solution, but the technical idealists want to dig their strength. A little closer to it." Jiang Guofei said.
Zhang Yongjie, senior product manager of Guangzhou Comma Smart Retail Co., Ltd., who was interviewed by Xiaobian, also mentioned that the cost of retrofitting traditional convenience stores through computer vision is far lower than that of sensors, and the application of sensors (such as RFID tags) The cost is very expensive and does not play a role in reducing labor costs. On the contrary, using machine vision to identify goods is not only accurate, but also inexpensive.
Wireless Sweatproof Earphone Bluetooth Headphones Bone Conduction Headset IP55 Waterproof
.
Companies registered capital of 35 million yuan, the end of 2014 the total assets of 48.69 million yuan, including fixed assets of 37.52 million yuan. The cothan another low way, so that more particles to have less than the activation energy of the reaction kinetic energy, thus speeding up the reaction rate. Enzyme as a catalyst, in itself is not consumed during the reaction, it does not affect the chemical equilibrium reactions. Positive enzyme catalysis, but also a negative catalytic effect, not only to accelerate the reaction rate, but also to reduce the reaction rate. And other non-living catalysts is different, having a high degree of specificity of enzyme, only a catalytic reaction or produce a particular specific configuration.test
.
Companies registered capital of 35 million yuan, the end of 2014 the total assets of 48.69 million yuan, including fixed assets of 37.52 million yuan. The cothan another low way, so that more particles to have less than the activation energy of the reaction kinetic energy, thus speeding up the reaction rate. Enzyme as a catalyst, in itself is not consumed during the reaction, it does not affect the chemical equilibrium reactions. Positive enzyme catalysis, but also a negative catalytic effect, not only to accelerate the reaction rate, but also to reduce the reaction rate. And other non-living catalysts is different, having a high degree of specificity of enzyme, only a catalytic reaction or produce a particular specific configuration.test
Wireless Sweatproof Earphon
Guangdong ganzhou , https://www.cn-gangdao.com