- About Us
- Results Released
- Application Scenarios
- Resource Center
Lab Platform Name
Unattended retail store lab platform
Beijing Vion Technology, Inc.
iSoftStone Information Technology (Group) Co., Ltd.
Since Amazon launched Amazon Go in 2016, checkout-free stores have drawn great attention from consumers and investors. In China, companies such as Alibaba, JD.com, Tianhong, and F5 Store have stepped into the unattended retail market. However, most solutions have only been implemented internally for trial use, and there are many unresolved issues.
In unattended retail, multiple AI technologies are used for recognition and monitoring of people and merchandise, and mobile payment methods are used for checkout. These require support from edge computing technologies and devices. Unattended retail stores are typical application scenarios of edge technologies in the new retail field, and an unattended retail store lab platform can integrate the ECC's resources to accelerate implementation of edge computing in this field.
The typical shopping process in an unattended retail store is as follows:
Going into a store
If a consumer goes into a store for the first time, they open an app on a mobile terminal, scan a code to obtain an entry code, sign clauses such as data usage agreement, privacy statement, and mobile payment withholding agreement, and then pass a gate to begin shopping.
Similar to shopping in a regular store, the consumer can pick up any available item in the unattended retail store or order food in the dining area. During the shopping process, an intelligent digital screen can automatically recognize the age and gender of the consumer and displays advertisements that may be of interest to the consumer. Cameras deployed in the store record positions where the consumer lingers as well as attributes of the consumer, and generate a heat map. The cameras monitoring goods shelves record the items chosen by the consumer.
Paying for the items
Before leaving the store, the consumer needs to pay for the desired items on an automatic payment machine.
Additionally, unattended retail stores allow consumers to pick up the items bought online by presenting a proof of purchase.
Unattended retail stores can also push advertisements to passers-by based on their attributes, which are detected by the cameras pointing away from the stores.
Unattended retail store deployment solution
Unattended retail store deployment involves synergy between the edge and cloud. Data collected by devices in stores is transmitted to the cloud for further analysis, and statistics about each consumer's shopping records and interests are generated and analyzed. Such information enables stores to provide better user experience.
Architecture of an E2E unattended store system
The lab platform applies AI to retail. The following outcomes are expected:
Person recognition and data collection using AI
As AI's application in facial recognition and motion recognition progresses, AI technologies can be used to enable automatic consumer recognition and data collection.
Automatic goods recognition and tracking
AI and radio-frequency identification (RFID) can be used for automatic goods recognition and tracking, which enables automatic collection of data about goods and simplifies the payment process.
There are currently two types of unattended stores: those using computer vision and those using robotic arms and RFID. Amazon Go, Tao Cafe, and Take Go are of the first type, using advanced technologies, including computer vision, deep learning, sensor convergence, convolutional neural network (CNN), and biometric recognition. 7-Eleven, Lawson stores where unattended checkout counter services are introduced, and Bingo Box are of the other type, relying on robotic arms and RFID. All these stores' solutions involve edge computing. To date, the transaction volume in the unattended retail market in China has reached CNY38.94 billion. ASKCI predicts that it will be increased to CNY1800 billion in 2022, by which time up to 245 million people will use unattended retail store services. Therefore, the application of edge computing in unattended retail has a promising future.
The retail industry in China rapidly develops under support from the government. In July 2017, the State Council issued a development plan for the new generation of AI. AI has become a part of China's national strategies, and the development of AI will likely promote the retail industry and product upgrades.