How to make the supermarket shopping more efficient?
Question:

 

PBL Design:How to make the supermarket shopping more efficient?

 

Introduction and Problem

    It’s not uncommon in our daily life to see people wait in a long line in supermarkets to check out. Sometimes you go shopping with joyful mood on weekends but get annoyed by the long queues. Sometimes you just want to grab a sandwich but get stuck in the crowds at the check-out counter. To deal with the low efficiency of the traditional supermarket checking out process, unmanned systems emerge as the time require. However, the present solutions roughly are all limited by technology defects or high cost. Therefore, you are supposed to figure out one way to solve the problem with mature technology and reasonable cost so as to promote it widely. The future of supermarket shopping is in your hand now.

 

Guidelines for students

    The students are divided into groups of 4-5 people. The task is divided into the following stages.

1. The first stage: identify a problem and form a preliminary idea (duration: 5 days)

    Students need to find a problem from real world. Each group member put forward a solution and then choose the best idea by discussion. Through this process, students can develop their innovation ability and critical thinking ability.

2. The second stage: do market research (duration: 5 days)

    Students need to do market research to know whether there are similar products and the market share of similar products. Market research can be conducted in three ways. One is to find relevant materials and data on the Internet, the second is field research, and the third is questionnaire survey. Through market research, students can understand the competitiveness of their products in the market and estimate the market share.

3. The third stage: build a model (duration: 10 days)

    Students need to draw or build a model of the product Including the specific structure of the product. Through this process, students can improve their hands-on ability and creativity.

4. The fourth stage: production (duration: 20 days)

    Students need to make products according to their own designs. In this process, there may be some errors, students need to continue to improve.

5. The fifth stage: Product presentation and group evaluation (duration: 1 day)

    The last stage is product presentation. Before presentation, students need to make a preliminary estimate of their products' market prospects and estimated profits. Then students should express their innovation and market value of their products to all other groups. Other groups act as venture capital firms and give decisions about whether to invest and how much to invest. This process can exercise students' evaluation ability and expression ability.

 

Additional Information

    Here are some technologies that may use.

1. Barcode or QR code scanning

    Barcode or QR code scanning utilizes the concept of "0" and "1" bitstreams that form the basis of the computer's internal logic in code compilation, using several geometric shapes corresponding to binary to represent information. 

                                

    These barcodes have been used throughout our lives, and we are already familiar with them. They are usually composed of black and white blocks with widely varying reflectance and only have the unique meaning.

    Our commonly used bar code reading devices can be divided into different reading principles:

(1) Linear CCD Imager

    In the CCD linear imager, an LED is used to illuminate the barcode, and tiny CCD sensors are aligned in a single row to read and decode the light reflected off linear barcodes. The device is like a camera that takes pictures having just a single row of pixels. Being linear, these devices could not read 2D barcodes.

    This kind of equipment is now often seen in supermarkets, libraries, etc.

(2) Image Reader

    The barcode image is taken by the area array CCD camera to analyze and decode the barcode image, and the one-dimensional barcode and all types of two-dimensional barcodes can be read. This principle is also commonly used by our mobile phones.

2. RFID

    Radio Frequency Identification (RFID) refers to a wireless system comprised of two components: tags and readers. The reader is a device that has one or more antennas that emit radio waves and receive signals back from the RFID tag.

    Tags, which use radio waves to communicate their identity and other information to nearby readers, can be passive or active. Passive RFID tags are powered by the reader and do not have a battery. Active RFID tags are powered by batteries.

    RFID tags can store a range of information from one serial number to several pages of data. Readers can be mobile so that they can be carried by hand, or they can be mounted on a post or overhead. Reader systems can also be built into the architecture of a cabinet, room, or building.

3. Image recognition

    The major steps in image recognition process are gather and organize data, build a predictive model and use it to recognize images.

(1) Gather and Organize Data

    The human eye perceives the image as a set of signals that are processed by the visual cortex in the brain. Image recognition tries to mimic this process. Computer perceives an image as either a raster or a vector image. Raster images are a sequence of pixels with discrete numerical values for colors while vector images are a set of color-annotated polygons, which is showed below.

    To analyze images the geometric encoding is transformed into constructs depicting physical features and objects. These constructs can then be logically analyzed by the computer. Organizing data involves classification and feature extraction. The first step in image classification is to simplify the image by extracting important information and leaving out the rest. For example, in the below image if you want to extract cat from the background you will notice a significant variation in RGB pixel values.

    However, by running an edge detector on the image we can simplify it. You can still easily discern the circular shape of the face and eyes in these edge images and so we can conclude that edge detection retains the essential information while throwing away non-essential information. Some well-known feature descriptor techniques are Haar-like features introduced by Viola and Jones, Histogram of Oriented Gradients (HOG), Scale-Invariant Feature Transform (SIFT), Speeded Up Robust Feature (SURF) etc.

(2) Build a Predictive Model

    After converting an image to a feature vector, now we should take this feature vector as input and outputs of a class label (e.g. cat or background/no-cat like below). we need to train our classification algorithm by showing thousands of cat and non-cat images. And then we use neural networks to build a predictive model. The neural network is a system of hardware and software similar to our brain to estimate functions that depend on the huge amount of unknown inputs. A neural network is an interconnected group of nodes, which would also require one learning algorithm. After tons of inputs, we will get a database which is big enough to build a predictive model.

 

(3) Recognize Images

    After completing the above two steps, it is very simple to identify the image. For example, if we want to judge whether the new picture is a cat, all we need to do is train a classifier that can take the measurements from the new test image and tells us about the closest match with a cat. The result of the classifier is the 'Cat' or 'Non-cat'.

 

Expectation & Requirements

1. Identify problems from real world and propose practical solutions.

2. Each group member actively participates in the discussion, expresses the idea and criticizes each other.

3. Form a complete plan and express ideas logically.

4. Divide the labor according to the specialty of the team members.

5. Conduct detailed market research.

6. Express your creativity and market value to venture capital firms.

 

 

 

※ Example 

Market Research

Currently there are two main products which are Unmanned Supermarket by Alibaba and Amazon Go.

 

(1) Unmanned Supermarket by Alibaba:

Customers should scan QR code to give payment permission of Alipay before they enter the store. After the customers select their commodities, they need to enter a paying house where the commodities will be recognized by RFID technology with tiny labels installed on them. When they leave the paying house, the payment process is finished on Alipay app. However, Unmanned Supermarket by Alibaba fails to be widely accepted because of its relatively high error rate, which is exactly the main problem of its core technology RFID. The cost of convenience is the loss of the store.

 

 

(2) Amazon Go:

Amazon Go is a chain of convenience stores in the United States operated by the online retailer Amazon. The stores are partly automated, with customers able to purchase products without being checked out by a cashier or using a self-checkout station. Customers must download the Amazon Go app for iOS and Android, which is linked to their Amazon account, before shopping at the store. The ceiling of the store has multiple cameras and store shelves have weight sensors, to detect which item(s) a customer took. If a customer takes an item off the shelf, it will be added to the person's virtual cart. If a customer places an item back on the shelf, it is also removed from the customer's virtual cart. However, the store concept uses several technologies, including computer vision, deep learning algorithms, and sensor fusion to automate much of the purchase, checkout, and payment steps associated with a retail transaction. It means the cost will be a huge problem if it is applied in supermarkets in a large scale.

 

 

 

 

Product Description and Model

 

Intelligent Shopping Cart:

Different from the traditional shopping cart, the intelligent one that we design has a camera on the holder and a sensor panel on the bottom. Before starting to purchase things, the customer need to scan the QR code on the cart with their cell phones. When the customer put an item in the cart, the camera will take a picture of it and upload it to customers cell phone. The app then will recognize the product and add it into the online shopping cart according to an updated database of the supermarket. The customer can see the price and other information of this item. If the item is fruit or vegetable that needs to be weighed, the camera will first recognize it and then let the app tell its price per unit. Afterwards, the sensor panel will weigh it and give its price in total. The sensor panel can also be used to help the camera to recognize whether the customer put in or take out any items. Moreover, customers can search any products online via the app and locate them in a very short time. When the customer finishes shopping and is about to leave, just click the PAY button and complete the payment process in one second without waiting in a long line at the check-out counter.

 

 

          

Cost Estimation

camera: $10   

sensor panel: $5   

 

Profit Pattern 

Compared to other solutions that require total redesign of the supermarket, this device only costs one more camera and a sensor panel than the traditional shopping cart.  The supermarket will no longer need the check-out counters and check-out staff. The saved space can be used for more shelves. For its high efficency, the supermarket can attract more customers and make more profits.

 

Technology Background

Image Recognition

The major steps in image recognition process are gather and organize data, build a predictive model and use it to recognize images.

 

 

Provide an answer by clicking on "Map" and then on "Add an Answer".



Results, Analysis and Discussion


Comments:
1.  Zhang Yujie (14.08.2019 pm 02.18.48)
real-life impact thinking but lacking effective analysis

First of all, the problem addressed affects a lot of people in real life and the group tried to solve the problem using various technologies. Good point!

However, the PBL design is disordered, the statement contains lots of grammatical mistakes and lacks logical consistency. At first reading, it is really confusing to know what the group is driving at.

Besides, the group already mentioned existing solutions to the problem, such as Alibaba and Amazon Go, but they failed to illustrate why their solution is any better than the existing ones. For example, the Amazon Go pattern certainly requires fewer human efforts, has more funding from the conglomerate, and bigger customer base. Then in order to equal Amazon Go, group 5 should address any advantages they may have in pricing, accuracy or market reach. But they did not.

It would be better to include competitive landscape, market supply and demand analysis as well as cost-benefit analysis.

2.  Zhang Yujie (14.08.2019 pm 03.18.59)
Peer Assessment - Group 6

2 Solutions:

  1. Improvement over the example design-facial recognition

One improvement is using facial recognition. In this way, it can pay immediately depending on the people’s faces, like Alipay, and can save even more time.

Different from Group 5, we just put cameras at the entrance, exit and exhibition shelves, which can save more money and make it more accurate. More specifically, when people want to buy some things, then the cameras at the exhibition shelves will record and count the goods in the system. If people want to finish their shopping, they just need to see the cameras at the exit and pay by cash or credit cards.

 

Appendix: Python code for face recognition

import numpy as np

faceCascade = cv2.CascadeClassifier(r'D:Program Files (x86)PythonPython37Libsite-packagescv2datahaarcascade_frontalface_default.xml')

cap = cv2.VideoCapture(0)

ok = True

while ok:

ok, img = cap.read()

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

faces = faceCascade.detectMultiScale(

gray,

scaleFactor=1.2,

minNeighbors=5,

minSize=(32, 32)

)

for (x, y, w, h) in faces:

cv2.rectangle(img, (x, y), (x+w, y+h), (255, 0, 0), 2)

cv2.imshow('video', img)

k = cv2.waitKey(1)

if k == 27:

break

cap.release()

cv2.destroyAllWindows()

 

2. Spread of e-commerce might eliminate the waiting problem altogether

It is said that the person that beats you can not only be your competitors, but also be someone from outside your industry, just like the automobile industry basically sentenced the transportation on horsebacks to death.

With the development of e-commerce, logistics and the change in consumer behavior, chances are that there will be fewer people actually being in a shop or a supermarket to get what they want, but instead they will order something online and have it delivered.

In this new shopping pattern, the supplier no longer needs to rent big space in a fancy mall, hire a lot of staff to work between shelves and counters, but instead the supplier can rent a big garage house in the suburban areas to save tremendously, employ robots to pick between commodities. Moreover, the consumers can order online anywhere anytime they want, and set delivery time and requirements.

Therefore, the problem addressed in the assignment may be put to an end in a surprising way. Since people are no longer in the supermarkets, of course there will be no lines!

3.  Hu Zhaoyang (14.08.2019 pm 09.00.47)
Comments by Zhaoyang Hu

 

I think your ideas are very great because of the embarrassment of crowd in shopping. Your video and graphics are well made, which makes people easy to understand. Although the ideas may have been occurred, your ideas presented include some improvements. For example, cameras on the shopping carts can record the goods and the value of them you choose. However, cameras on the shopping carts may cost a lot of money, which will make the profit become less.

 

In addition, I think that there can be an important improvement---face recognition. Utilizing some easy and basic codes can make this happen. Face recognition can elevate the level of security and accuracy of this paying system, because we can put cameras at both entrance and exit to record people who come to shop. Another improvement is that you should pay attention to protect QR codes from breaking. If QR codes are broken, then the whole system of convenient shopping will be paralyzed.

 

In short, your ideas are very creative in daily lives. If this new mode of shopping can be spread all over the world, it will save billions of people’s time. However, some details such as the cost of this new mode of shopping should be discussed more specifically.

  

4.  Peng Xiaoxuan (14.08.2019 pm 10.31.41)
Comment by Xiaoxuan Peng

I think the Assignment of Group 5 is generally very innovative, because the kind of inconvenience they brought up that people wait for long time to check out almost happened to everybody before. The product they designed is easy to implement, and it can greatly reduce the cost, since all they need are just the camera, the weight sensor, and the APP.

However, I think this product doesn’t have a long-term development prospect. As we all know, the retail method is changing nowadays, for E-commerce is rising rapidly and the traditional retail sales are getting less and less profit. In my opinion, it’s the change of selling method that matters.

5.  Han Zhe (15.08.2019 am 12.11.56)
Comments by Zhe Han

Your idea is good and relevant to our daily life but lacks innovation to some extent. 

First, it is important to note that there are some products already in the market which can save us time. For example, some Apps in China can send the goods to you in 1 hour. You even don't have to go outside.

Second, we can buy many goods from different destinations according to our needs. We can buy drinks and snacks in convenience stores or vending machines. And we can buy fresh meat and vegetables in farmers' markets. 

Third, in my opinion, there are other solutions to this problem. If we want to go shopping in real shops, we can select the good we need on the App. Finally, all the goods will be automatic collected in the warehouse and be delivered to us. Besides, we can have a VR shopping at home.



History of edits
Edited BY: Yanzhou Pan Edit Date: 2019-08-14 14:53:05
Edited BY: Yanzhou Pan Edit Date: 2019-08-14 14:13:48
Edited BY: Yanzhou Pan Edit Date: 2019-08-14 13:48:52
Edited BY: Yanzhou Pan Edit Date: 2019-08-14 11:55:00
Edited BY: Yanzhou Pan Edit Date: 2019-08-14 11:51:48
Edited BY: Yanzhou Pan Edit Date: 2019-08-14 11:46:46