Comprehensive Analysis, Design, and Development of Amazon Go System

This project delves into a detailed analysis of Amazon Go, the innovative cashier-less store system. It begins with an introduction to Amazon Go, outlining its objectives, rationale, and the benefits it offers to customers. The analysis section examines the problems and issues associated with the system,...Read More

P a g e 1 | 14
Table of Contents
Task # 1 Analysis and Identification …………………………………………………….1
1.1 Brief introduction…………………………………………………………………….....2
1.1.1 Objective and rationale………………………………………………………..….2
1.2 Project identification and Analysis………………………………………………….3
1.2.1 Explanation of Problems and issues………………………….……………...3
1.2.2 Analysis on methodology of Amazon go………………………………..…...3
1.2.3 Technology of Amazon go.……………………………………………………4
1.3 Analysis on the existing system…………………………………………………4
1.4 Suitable approach for development…………………………….……………….5
Task # 2 Design……………………………………………………………………………….4
2.1 Design # 1………………………………………………………………………………..5
2.2 Design # 2……………………………………………………………………………..6
Task # 3 Development……………………………………………………………………….4
3.1 Development of prototype……………………………………………………………...5
3.2 Documentation of prototype………………………………………………………6
Task # 4 Implementation…………………………………………………………………….4
4.1 Implementation…………………………………………………………………………..5
4.2 Implementation challenges……………………………………………………….6
Abbreviations…………………………………………………………………………………
Appendix 1 Definitions of table……………..………………………………………..
Appendix 2 Definitions of query……………….……………………………………..
Appendix 3 Artefacts of prototype development……………………………………
Appendix 4 Data and tests………………………………..………………………….
References………………………………………………………………………………………

P a g e 2 | 14
1 Analysis and Identification:
1.1 Brief Introduction:
Everyone knows about the Amazon over the world wide. Amazon is the online sell or
purchasing store and application. Where people buy and sell their products and most of
the audience rate occupied the Amazon. And the Amazon go is the part of the Amazon
store where people go for grocery and goods. (Lauren, 2018) Amazon go is the most
advance development form for people now a days. Amazon go name show that the go
means grocery and “go for walk out shopping”. Amazon go is the online means
Ecommerce store or application where allows the people buy the goods and items in
physical stores deprived of waiting in lines for check out and registration for check out.
(Lauren, 2018)
1.1.1 Objective of Amazon go Information System and rationale:
The objective of Amazon go store is the benefit for people to go for walk out shopping
where people go for free up the check out registration and without waiting in lines. This
technology of Amazon, is based on “just walkout technology” and uses the different
smart type of sensors. Amazon go has divided into two parts. First part is base on the
physical store and the second part is online mobile app connectivity. The mobile
application is determining the customer changes and the items in return to the shelf.
This new system is time consuming and better than existing line waiting system of
checkout.
There are different stakeholders are associated with this new information system. First
one is IT director, this person sees that the app working or the resolving the errors
through the IT workers or team. The second person is finance director who handle the
fraud system and causes or consequences of the store. Customer care employees who
see the customers movement and availability at the time in the store and resolve the
customers issues and problems. According to (Lea 2019) the main shareholders of the
Amazon company are Jeff Bezos, Vanguard, Mac Kenzie and research company.

P a g e 3 | 14
1.2 Identification and Analysis:
1.2.1 Explanation of Pressure and Problem:
The main problems and issues of this project are when people enter into the store then
mobile application is compulsory online or not the user then enter into the store and if
any user enter into the store without registration, then Administrative team check these
issues are active all time? And this project is highly risky because many types of
scammers are live in society. This is just the walk out shopping? And this type of
technology is work in the crowds or across distances, Amazon has many other issues
and problems faced in previous years.
1.2.2 Analysis on methodology of Amazon go:
Amazon is introducing the high type of technology retail in the location called Amazon
Go, recently in a private beta testing in Seattle and scheduled to open to the publica
early next year. The big point of advantage and selling is no check out line.
(Todd, 2016) Eventually, Amazon go works with the Artificial intelligence (AI).
According to the amazon go using methodology and generate the benefits for peoples
where everyone has futuristics convenience store and everyone simply go for walk in
and buy what you want according to the needs and walk back out. Within minutes or
seconds, you can receive your receipt on your amazon account through mobile app.
Amazon go use the technology and it’s an entire work of machine learning algorithms
and innovation.

P a g e 4 | 14
1.2.3 Technology of “Just Walk out”:
The core thing of the Amazon go store technology is whole system based on the
Machine learning. Machine learning is used to flawlessly path and approximation the
purpose of everyone in the store. (Ryan, 2019) Amazon departed into an amazing level
of feature on their implementation of this technology. Amazon not display the actual
neural architectures for their models. Amazon only show the particular issues those
separate models resolve and how joint to build a full resolution. (Ryan, 2019)
Through Amazon go app users or customers simply register them into the store and
using the app choose the items or grocery, customers or users are free to exit the store
and the payment criteria handle the app account. Where customer or user make
account and connect to the bank account then automatically deduct the amount of you
choose items through Amazon go app.
Fig 1 Architecture or framework of Amazon go

P a g e 5 | 14
1.3 Analysis on the existing system:
Why adoption of cashier less will be slow:
In the new technology, one complexity is discovered in the Amazon go system its
tracking technology. Tracking technology originally couldn’t grip more than two dozen
shoppers in the store. Another big issue is retailers installing a cashier less decision is
that customer enhance on the sales take a dump and customers are not attracted to
grip on an extra item or any other instinct buy given there is no stand in to checkout line.
Using the machine learning algorithms of AI risk assessment not occurred because the
algorithms results is always accurate or correct according to the needs and
requirements so due to the machine learning computer vision mean the automatic
inventory and less theft ratio and more secure or space available to increase revenue
per square meter. (Adrien, 2019) Using the machine learning algorithms gives us the
admittance of additional data and users are not only linked over the account to track
and name what they purchase, but also graphically tracked around to track what they do
not purchase. (Adrien, 2019) Basically, the machine learning depends on packaging to
recognize products. Particularly, a mixture of printed outlines to make recognition
easier. (Adrien, 2019)
1.4 Suitable approach for development:
The suitable approach for the development is already on the success point right now
because amazon use the best machine learning algorithms for amazon go convenience
store and the competitors follow them but not successful then amazon go stores.
Here is the suitable approach for the cashier less retail system in Amazon go:
The customers opening of Amazon go store, which technology no cashiers, is being
delayed due to some Amazon’s technical issues. The very big challenge is that sensor-
based check out structure is not working on well basis because when there are closely
two dozen shoppers available in the store and is not following and tracking items as
expected. The main problem is occurred in the system in tracking more than 20
customers at a time. Store of Amazon go situated in Seattle according to the customers
where uses cameras, algorithms and sensors to watch the users and track what they

P a g e 6 | 14
pick up. But Amazon has run into the issues more than twenty customer in the store at
one time, according to the customers the difficulty of keeping tabs on an item if it has
been moved from its particular spot on the shelf.
Although Amazon Go stores and their competitors are grounded on the similar
investigational ambient intelligence system, the technology overdue their systems differ
change. The Amazon Go’s monitoring procedure is managed by picture recognition,
machine learning and artificial intelligence. With the sensors and shelf cameras track
products detached or swapped, and a 3D picture of every user is constructed and
tracked. Then there’s a backend scheme which processes the transaction mechanically
for a flat and unified shopping skill. Alibaba’s Hema, on the other point, using the facial
recognition at self-checkout though the South Korean Lotte using biometric
confirmation.
Amazon go system is logging the stuffs as the shopper drives along, which removes the
need to go over a traditional check-out line. When users leaving the amazon store over
a “transition area,” the amazon system senses that they’re exit, adds up the stuffs and
charges their Amazon user account.

P a g e 7 | 14
TASK # 2
Taking analyzed on the present situation in the earlier segment, this segment delivers
the design of the system of Amazon go and specific information requirements on it.
Process models, and technical solution and database design.
2.1 Design of Amazon system ERD:
2.1.1 ERD design of the new Amazon go system:
Before go to DFD diagram firstly understand the ERD diagram of the system because
ERD is first step of understanding the whole system and its working criteria. ERD
diagram include the major aspects and features of entity relationship diagram and its
process model, this segment of the report analyzed and Display the design of the new
Amazon system based on more customer at a time in the store.
The ERD in (fig 2) main entities involve are Administrative staff (IT) or casual staff in the
store for check the activity of the customers and check the stock availability and the
next entity is customer, and Amazon go store. The first entity Administrative staff or IT
staff check through the camera customers movement or activities and check the
sensors working on the right path or not. The next entity is casual staff check the
movement of customers physically because if IT staff checking the fault through
surveillance camera then ready for action on urgent basis for the fault correction or
fraud customers. The third entity has been involved is customer. Customers are
basically many types and every customer must be register or login through the Amazon
go app of the store then enter into the Amazon go store. The last entity of the system is
Amazon go store and its attributes are check the available stock and location of the
store send to the warehouse then generate the daily report of the store.

P a g e 8 | 14
2.2 Design of Amazon go store DFD;
2.2.1 DFD of the Amazon go new system:
DFD has three level of a system. Now in this design section I have to discuss about the
DFD level 1. In this DFD level shows four important entities or changes in the system
where many people buy things at a time. DFD show that the relation between entities
and system. DFD (fig 2) show that the new system about the Amazon go store where
more customers buy items or grocery at a time.

P a g e 9 | 14
2.2.2 Design of DFD level 2:
In this level of DFD define the system in details and elaborate the DFD level 1. In this
diagram, see that the working of whole system and how customer select the items and
amazon app add the items into the cart. And then generate the receipt and send to the
Amazon go account and customer account and then save into the transaction database.
Fig 2 DFD level 1 of Amazon go store

P a g e 10 | 14

P a g e 11 | 14
2.2.3 Information requirement specification:
The requirements of information and specification of the system can be designed by
using DFD diagram define the input and capture process, storage and communication,
and output of the system.
Input specification:

P a g e 12 | 14
Data will be captured either the User-interface design through database system. Data
enter of the typical type by users through the system UI of app and create the account
through the Amazon go app.
Specification of process:
All process system is depending upon the functionality of the system. Generally, the
system specifies the given input through the app by users are correct or not. Then
further process of the system goes on.
Specification of storage:
Primarily data will be stored and retrieved from the rational database. All the data stored
into the database after the completion of processes and save data for latter when any
customer use or see history of the transactions easily.
Specification of output:
Output of the new information system depend upon the user, but approximately different
types of output can be well-thought-out. First one is transaction of items specific.
2.3 Design of Technical solution:
The technical solution of the Amazon go store or information system is “just walk out
technology” follow these steps how to work Amazon go store.
1. Customer registration: Store can link their Amazon customer account.
2. Customer’s location track: System can relate to the customer data and the action
taken place.
3. Detect an item picked up: System can add items to the virtual cart of shopping of
the customer’s virtual cart of shopping.
4. Detect an item put back onto the shelf: System can delete products from the
customer’s virtual cart of shopping.
5. Detect exit the store of customer: customer’s working transaction can be finish
and complete or leave the store.

P a g e 13 | 14
1 Registration of customer:
Firstly, users download the Amazon go app to their phone, which is not part of the
Amazon go app. At the entrance of store, they have to scan the QR code on their app to
the door, which nearly looks like some sort of a subway entrance.
2 Customer’s location tracking:
There are hundreds of cameras mounted on the ceiling. They are RGB cameras for
tracking individual customers. Amazon has mentioned that their go stores don’t use any
facial recognition technology. Instead, these cameras detect each customer’s general
profile and track individuals with motion detection. The cameras relate a customer exit
camera A and picks up the same customer entering camera B. The accuracy of tracking
is augmented by use of separate depth sensing cameras.
3. Detect an item that picked up and back:
That is the exclusive feature of the Amazon go and is characterized in its store design.
Every shelf has a weight sensor that recognizes the particular weight of each item.
When an item is picked up, the sensor can tell accurately which shelf the item if from.
Likewise, the sensor tracks when the item with the same weight is put back onto the
shelf. The CPU central processing unit narrates the information about every customer’s
address and the movements taken place on every shelf. Because of this system design,
every shelf has clear guides separating every row, and they are additional spacious
associated to regular grocery stores. The store always looks neat and well planned,
because objects need to be placed exactly, and space helps exactly track customers.
5 Detect exit the store of customer:
Customers don’t have to scan the QR code to exit like they do when they enter. In-store
tracking detects when they leave the store. When I walked out from the store, I was
curious if the store successfully detected items that my wife picked up. In fact, it took
about 5 minutes after leaving the store to receive my receipt and see any updates on the
app. I am not sure if this was by design, but I hope Amazon Go app had updated my
virtual shopping cart while I was in the store.

P a g e 14 | 14