Cloud Computing and its Based IoT Applications in Haidilao’s Unmanned Restaurants By: Xuying Wen | ME20192334 Daniel Wang | MA20170148 Yuzhou Huang | ME20192385 In the near future, computing power will become something like electricity, which can be flexibly and efficiently distributed to each end user via massive network. Just imagine, each company can launch its own information center, manage its own business data and conduct on-line transactions with numerous customers as long as it has certain terminal displays! The exciting fact is this forecast has been gradually realized by a great technology — cloud computing. What is cloud computing? It’s a very complex technology, and I’m trying to explain it in a simple way. When a customer wants to buy a computer to deal with personal documents or play games, he/she will pay attention to a computer’s parameters like CPU, memory, etc. Similarly, a company will purchase servers with proper parameters to establish its information center. By the end of the last century, companies usually purchased physical servers to establish their information centers, however, there are several pain points for companies to buy physical servers: 1) the procurement cycle is quite long, companies usually got those physical servers weeks after they bought them; 2) physical servers are usually very expensive for their extremely powerful CPUs, memories and other parameters, however, it was quite common that a company had to buy an extra sever even if it only needs to upgrade the performance of its current server a little bit more; 3) the maintenance of all those distributed servers in different geographical locations was very boring and time-consuming. P a g e 1 | 14 When we have stepped into a new century, cloud computing technology, which was first launched and widely applied by Amazon, has been developing significantly. Could solve all the above problems. In a nutshell, the realization of cloud computing can be divided into two steps: 1) utilizing virtual machine technology to assemble the computing power of hundreds of distributed physical servers into an enormous resource pool; 2) using unified tools to allocate resources (e.g. computing power) to each individual user automatically and immediately. In addition, a user can enjoy the convenience of cloud computing as long as he/she can log onto the Internet. Amazon, Microsoft, Alibaba, Google, and Tencent are some of the most famous cloud computing providers worldwide. Nowadays, if a company wants to launch its own information center, it can buy the computing power from Alibaba and get the computing power immediately, instead of obtaining the physical server weeks later. It can quantify the performance of the information center it needs, for example, 1000 standard CPUs, 100G memory plus 10G bandwidth; if it’s going to upgrade the performance a little bit more, instead of purchasing an extra expensive physical server, it is feasible to buy one more CPU or 1G more bandwidth, which is much more flexible and economical. In addition, the company does not need to deal with most of that trivial maintenance of its ‘online’ information center, because Alibaba, the cloud computing vendor, will help do it. Therefore, cloud computing enables companies to access massive computing power immediately, obtain the number of computing power according to its exact need a certain point, and save a lot of maintenance cost. Cloud computing technology has been widely applied to IoT (Internet of Things) systems. Generally speaking, an IoT system consists of four layers as follows: Figure 1: The framework of IoT system Sensor Network Platform Application P a g e 2 | 14 In the Sensor layer, data are collected by numerous sensors installed on-field. Typical collected data are voltage, currency, temperature, pressure and other physical information. These data are transmitted to backend through the Network layer, which is data transmission devices based on protocols like Wifi, Zigbee and NB-IoT. In the Platform layer, specific algorithms are used to analyze those massive data collected on-field. These algorithms are usually industry-specific, for instance, in the logistics industry, one important algorithm is to realize location-based service; in the e-commerce industry, a widely-applied algorithm is to analyze customer behaviors and predict their potential consuming needs. Finally, the Application layer is closely tied to specific functions like visualizing analyzed results or providing feedback signals to control those on-field electrical devices. By the way, the most common application scenario of IoT is remote monitoring & control. Within an IoT application, the data analysis task is extremely heavy. There are several reasons: 1) hundreds of thousands of sensors are usually used in an IoT system, and each sensor continuously generates data; 2) in addition to those collected on-field physical information, time, address, device, and other information are usually integrated and transmitted to the backend for analyzing. Therefore, the Platform layer of IoT often takes advantage of cloud computing technology, which has enormous and elastic computing power, to deal with those massive data. On the other hand, I also would refer to edge computing as a perfect supplement for cloud computing. Edge computing is conducted in the Sensor layer for application scenarios with a high real-time requirement such as unmanned driving. In fact, some intelligent sensors have an edge computing function to pre-analyze collected data or communicate with other sensors nearby. To summarize, cloud computing is mainly responsible for complex data analysis remotely while edge computing mainly tackles light-weight analysis locally for applications with high real-time P a g e 3 | 14 requirements. The combination of the two can efficiently deal with big data tasks within an IoT system. Figure 2: Cloud computing & IoT applications along business chain Sales & Marketing Logistics Manufacturing Frontend Backend • E-commence • RFID scanning and storage • Smart equipment tracking • Online payment • Automatic guided vehicles • Flexible production line • Advertising effectiveness evaluation • GPS position of fleets • Accident predicting and warning • … • … • … IoT based on cloud computing has been widely applied in business operations and will continue exploring more application scenarios. In the past decades, with the rapid popularity of smartphones and great upgrading of IT infrastructure, cloud computing & IoT have significantly change the operations of sales and marketing. One interesting application is the evaluation of digital advertising effectiveness: the embedded cameras in the ad post can track passers’ eye and understand which parts of the advertising attract them most based on backend complex algorithms, thus to effectively modify the advertising contents. IoT also have improved the logistics efficiency a lot: 1) in the warehouse, RFID scanning and machine vision can identify, locate and track items efficiently; AGVs move cargo automatically and save a lot of labor costs. 2) during the transportation, IoT technology helps to monitor the status of cargos and fleets, give driving feedback to the drivers and optimize the driving routes. Furthermore, we see a trend that IoT & cloud computing are gradually sinking down along the business chain, from the frontend to the backend, and will have more potential in dealing with the flexible production line, smart equipment tracking and other scenarios in the manufacturing process. Recently, the pilot unmanned restaurant launched by Haidilao is a great example of IoT P a g e 4 | 14 based on cloud computing, which covers all those parts from the front end to the backend (sales & marketing, logistics, and manufacturing). With the development of new technologies and consumption upgrading, big data and artificial intelligence (AI) have become the keywords of the food industry this year. On October 28, 2018, Haidilao, one of the leading Chinese hot pot restaurants, officially launched its first "unmanned restaurant" (or named smart restaurant) in Beijing, China, costing around 150 million Chinese Yuan (Huang, 2018). New technology has begun to disrupt the food and beverage industry. Haidilao restaurant can realize the operation by robots in most of the areas, including waiting for the table, raw material cleaning, dishes preparation, hot pot soup base preparation and food delivery. Before this, Haidilao has used big data and artificial intelligence many times to optimize its selection of siting and inventory management. For example, it has developed an AI platform with Ali Cloud technology to improve the selections on its website. The algorithm sifts through population density, business opportunities, nearby restaurants and nightlife spots (Huang, 2018). Many Haidilao restaurants open 24 hours a day based on the algorithm results in terms of cost and benefits. The AI platform also calculates centralized inventory management. Based on the in-depth learning programs regarding table turnover rate and consumption trends, Haidilao can meet consumer demand. In the meantime, it can monitor fresh food such as lamb imported from other areas. P a g e 5 | 14 In October 2018, it is the first time that Haidilao has realized automation of service, and replaced waiters and waitresses with robots for basic services. The main improvements consist of reducing basic staff, enhancing the consumer experience and increasing supervision and management. Reduce basic staff In terms of personnel