PRE2016 3 Groep17
Concept: Automated product scanning in supermarkets
Team members: Ava Swevels, Lisan Wolters, David Elshove, Vansh Kharbanda, Johan Kon & Thomas van de Wiel
How to use this page
In this wiki, a weekly update will be given on the progress of the group. The total Wiki is therefore divided into weekly modules. At the end of the project, the conclusions and results will be given, as well as a discussing of the concept and an evaluation of the group process.
In week 9 you can find our final report in PDF. This report includes a general and concise overview of the research we done over the past eight weeks with regards to the Automatic Product Scanning System.
Note that the first week of the project, the team decided to investigate and research a street-level garbage sorting system. This had been changed to 'Automated product scanning in supermarkets' from week 2
Selection of idea
The first idea we had was to make an educational application in which we would make several short lectures for last year secondary school students. With the knowledge they gained, the could then make a working robot from a package of parts. By doing this, we would show the students how fun an engineering education can be, in order to make them seriously consider higher education in a technical institute. However, it proved to be hard to distinguish our idea from the existing "stichting techniekpromotie", which does pretty much the same thing.
We then thought of a a societal problem which really needs to be solved, such that we could think of a solution that could solve it. We came up with the following. Garbage sorting is a huge problem in society. Not sorting your garbage has a serious impact on the environment. Some team members indicated that especially students omit sorting their waste. What if we could solve this problem by making a bin that automatically sorts your waste, once you throw it in? Then we figured that this bin would become rather big because of all the sensors and the actual sorting system, and that most students would not have room for this. Furthermore, it would be an expensive solution for which the target audience probably wouldn't want to pay. But since students are mostly clustered in an area, we could scale up the idea to come across this problem. What if we make a garbage sorting system on street level, similar to already existing underground trash cans (), but in which you can throw all garbage at once if you did not sort it. This way we avoid high costs for the user, while maintaining the benefits for society. This idea got the groups approval and was thus selected for further investigation. The group decided that it wanted to do a study on the possible effects of such a system and to develop a working (scaled down) prototype of the system.
A literature study to similar existing systems is done, as well as a study to the possible societal impact of such a system. Furthermore a presentation is be prepared in order to get green light for the execution of the idea. To visualize the idea and to add convincing power, a 3D CAD-model is made. The model will also serve as a step up towards the prototype and to make sure all team members are talking about the same concept. A discussion will be given below.
This section reviews the state of the art and related work relevant to our research. A list of the current knowledge in the field of our research has been made. The used sources are listed below.
There already exist several automatic sorting trash cans, each has their own purpose and implementation. The first sorting bin we will discuss is the Auto-Trash from Jay Donovan. This was a technical craft project to sort between compostable items and recyclable items. They have used a Raspberry Pi module, which was equipped with a camera to detect and recognize each item placed on top of the can’s rotating top. Image recognition was used to distinguish the items, and then the item is sorted appropriately (Donovan, 2016) . Our product differs since it will be implemented on larger scale, namely on the level of the municipalities and not on the level of personal houses. Furthermore, our product will be able to sort the items in more different categories and it will also be able to sort several items at the same time. Therefore the user does not have to adapt to throw his trash one by one into the bin.
The second sorting bin is the R3D3 from the company Green Creative. This is a very small and compact bin which can be placed in public spaces and in work areas. This bin accepts one item at a time, then it detects and recognizes the item and then, after compressing the item, it is sorted appropriately. This bin accepts 3 sort of items, namely cans, disposable cups and plastic bottles. When the bin is almost full, the garbage collector will get an alert to empty the bin (Green Creative, 2017). Our product will have different features than this bin, and therefor will have an added value. Our product will be on a larger scale, as stated before and our bin can also sort other materials. Furthermore our product will also be able to compress the waste and it can also alert the garbage collectors when the storage space is almost full. This way they will never have to make a trip to collect a half-full garbage storage unit, which saves money.
The last sorting bin differs from the first and second one in terms of scale. This bin is developed by Sintef, a Norwegian research team. They have developed a sorting machine for waste on a large scale, namely on factory-level (Sintef). The sorting machines use spectography and infrared light to analyse the materials in a product. Then the items are identified and sorted into various divisions (Sciencenordic). Our sorting machine differs since we won’t implement our machine at this level, but on a lower level, namely in the municipalities. This is more convenient, since a lot of cities already have underground waste container, thus the sorting can take place in these containers. Therefore transport will be reduced, since the waste does not first have to be transported to a factory to sort it, and then transported to the factory where the materials will be recycled.
Evaluation concept with repect to USE
Our automated sorting trash cans will have impact on users, society and enterprise. Not every household will have an own automated sorting trash can. It will only be place on a limited amount of places in the municipalities.
For some of the users of the product it will make sorting easier. Nowadays we have to sort our trash in a lot of different containers. These containers can be replaced by only one container. If the trash bag is full the trash will be thrown into an automated sorting trash can. This will reduced the amount of containers households will have as well as the amount of effort needed. While we do not need to know in which container we need to sort which material. In some municipalities there is a penalty for sorting wrong. If a person has thrown something in the wrong bin, even without noticing. This person can get a penalty(Tilburg). This problem will get solved for the user, while he/she cannot throw trash in a wrong container. “ “It’s really beneficial for individuals to sort their trash at home so it can be recycled. This enables them to contribute to a better environmental utilisation of the resources in the trash,” says Political Advisor Audun Garberg at the Ministry of the Environment. ” (Sciencenordic). This refers to an impact which can happen, but must be avoided as much as possible.
The society will have more assurance that trash is sorted correctly. It is humanly to make a mistake when you try to sort your waste yourself. Besides there are still a lot of people who do not sort their trash. All the thrash which is not sorted now, will be sorted in the future, which is better for the environment while more products can be reused/recycled (Milieucentraal). In the beginning this invention will be an investment for society, but this will later earn itself back. While there is less transport needed. With this automated sorting trash cans not all household will have their own container, but only limited places in the municipalities will have these. This will shorten the route of the garbage trucks. More materials can be reused as well while all trash will be sorted correct. This increases the amount of materials we can reuse instead of producing them again, we can save money as well (Tilburg).
For enterprise this product will change the work for companies which collect the garbage. Garbage containers will not be placed at all houses, but only at one point of the street. There will be sensors included, with which can be measured whether the garbage container must be emptied. This will change the route the garbage men need to take to collect the garbage. It will also change how many times they need to empty some of the garbage containers. This can be checked every day.
Presentation of ideas
The presentation will be given on Thursday 16/02 between 13:45 and 17:30 in the Metaforum building, room MF08.
Some pictures of the CAD-mockup can be found under the links below.
Side 1: 
Side 2: 
Donovan, J. (n.d.). Auto-Trash sorts garbage automatically at the TechCrunch Disrupt Hackathon. Retrieved February 09, 2017, from https://techcrunch.com/2016/09/13/auto-trash-sorts-garbage-automatically-at-the-techcrunch-disrupt-hackathon/
Green Creative, R3D3 smart and connected sorting bin. (n.d.). Retrieved February 09, 2017, from http://www.green-creative.com/en/r3d3-sorting-bin
Sintef. (n.d.). Automatic sorting of waste for recycling. Retrieved February 09, 2017, from https://www.sintef.no/en/projects/automatic-sorting-of-waste-for-recycling-/
Sciencenordic, Machines are better than people at sorting household trash. (n.d.). Retrieved February 09, 2017, from http://sciencenordic.com/machines-are-better-people-sorting-household-trash
Tilburg, controle afvalscheiding, retrieved February 12, 2017 from https://www.tilburg.nl/inwoners/afval/controle-afvalscheiding/
Milieucentraal, afval verwerken, retrieved February 12, 2017 from https://www.milieucentraal.nl/minder-afval/afval-scheiden-en-recyclen/afval-verwerken/
The group has decided to switch to researh and the making of a prototype regarding a new barcode scanning technology that could reduce the waiting time at supermarkets and possible replace checkout personnel. We will start by giving an introduction of an alternative technology that could be used to replace barcode scanning, indicating why we think it will not be used the coming years. This is followed by research on the current state-of-the-art. Lastly, the user and enterprise questionnaires that will be used to gather data on this technology are presented. Note that this week the focus was on making and dividing a planning, which explains some delay on content. The planning can be found at . The project plan contains goals, sub goals and concrete activities with deadlines and a task distribution. A Gantt Chart has also been included to show the time span of the activities within the project, indicating important milestones (black), interim deadlines (yellow) and project deadlines (red).
RFID in supermarkets as a competitive technology, alternative to EAN-scanning
Barcodes as we know them have been used in supermarkets since 1974 already.  In all these years, the technology behind then has not changed significantly. Over the years, lots of new technologies have emerged. So why would it be a good idea to improve this old barcode system instead of revolutionizing the supermarket checkout system? To investigate this, we will look into one of the current technologies which could be used for this purpose: RFID technology. In this article, the general principle of RFID tags is explained. After that, the advantages and disadvantages of RFID technology are listed. Finally a conclusion is drawn regarding future possibilities for barcode and RFID technologies in the supermarket scene.
First of all, what is RFID technology? RFID of Radio-Frequency Identification. It is a general term used for small electronic devices consisting of a small chip and an antenna. The device can carry small amounts of data.  The basic working principle is the following. When the RFID-chip detects a field of a scanning antenna (one that asks for information), it gets activated. The RFID-chip then sends its information to the antenna, which now knows all details about the chip that was detected.  Examples of fields where RFID technology is currently used are banking and postal services. More and more banks equip their credit cards with RFID technology, in order to enable users to pay small amounts of money with just moving their card near an RFID scanner.  Postal delivery companies use them for track and trace of their packages to give customers real-time information on the status of their order.  With RFID-chips becoming increasingly cheap (currently one to several cents ), the door is opened for many more daily life applications.
A logical extension to this contactless way of tracking items trough small chips, is scanning of products in supermarkets. The main advantage of RFID tags over barcode technology is that using the former enables faster checkout rates. Customers can walk through RFID gates instead of having to put every item on a conveyor belt, having them scanned by a human, putting them in their shopping cart again. It would substantially reduce time needed to shop and would make the overall experience a lot more pleasant.  Supermarkets would benefit from this technology in lots of other ways as well. The biggest cause of shrinkage in supermarkets is caused by internal theft. RFID technology would make it possible to pinpoint when and where items disappeared, making it easier to prevent similar events in the future. There are already examples of decreased shrinkage rates in supermarkets using RFID tags. Secondly, theft by customers would be more difficult because of the previously mentioned checkout gates. Another example of the benefits of RFID tags is the possibility to track where in the supermarket the products are located. Have they been put in the shelves yet? How many items are still stocked? What is the expiration date of certain products? This extra knowledge would enable supermarkets to stock, move and sell their products more efficiently, thereby reducing financial losses due to excess stock or spoilage of products. It would also make it possible to automatically send an order to an employee to restock a shelve once this is necessary. Furthermore, automatic orders for restocking the stockroom could be done.  Besides faster checkout, customers comfort could be improved by using other features of RFID technology. For example, customers can check whether the product they want to buy is still available at the supermarket, so that he knows whether he needs to go or not. Supermarkets could equip shopping carts with a system that automatically registers anything you put in or take out of your shopping cart and display the price of your items real time (‘smart carts’). Estimated arrival times of out-of-stock products can easily be communicated with customers. Customers could be informed of their optimal shopping route based on their online made shopping list and the information the store has on the location of its products. Additional product information could be given, using RFID tags and smart carts. 
The list of advantages this new technology offers is inexhaustible. So why are we not all shopping like this already? First of all there is the price of producing the tags. A price of several cents might not seem much, multiply this number with the amount of products in your weekly shopping cart and ask yourself whether you still think it is not that much. Fact is that a lot of supermarkets are just not able to bear this small added cost without disappointing customers and losing incomes.  A second issue is the cost of changing the entire checkout system. Changing to an RFID-based system would require massive changes in supermarket infrastructure, which go hand in hand with high costs. In the ever more competitive market of today, a supermarket simply does not have the possibility to change the entire system without increasing the cost of their products.  Thirdly, it is impossible that all manufacturers of products that are in the supermarket change to RFID-technology at the same time. There will be a transition period in which both barcodes and RFID-tags will be used. This would pose big problems for checkout system. Supermarkets would be forced to use conventional barcodes until all products are equipped with RFID-tags. This would create a situation in which the store pays for the RFID-tags, without being able to use them to their fullest extent.  Another issue is the current technological state of RFID-tags. Products could be missed at the checkout when multiple signals overlap, since the reader is only able to accept one signal at a time. Also, products in the middle of your shopping cart could be missed at the checkout because the RFID-signal just isn’t strong enough. Smart carts are one way to avoid the last problem, but nonetheless, lots of issues remain. 
We have seen many ways in which both customers and supermarkets could benefit from RFID technology: faster checkouts, tracking of stock, reducing theft and spoilage of products, making the overall shopping experience more pleasant. However, there remain some issues that make it just not feasible at the moment to make the transition to RFID technology as a replacement of barcodes. Mentioned issues included the both price of the tags and the cost of changing the entire supermarket infrastructure. Add a costly transition period and current technical issues to that and the conclusion is clear: RFID technology is not (yet) able to revolutionize the shopping experience. The coming years, barcodes are still expected to play a huge role in the checkout line of supermarkets. It is therefore not wasted effort to develop a new scanning system that makes faster checkout possible with the current barcode system. Au contraire: it would overcome huge investment issues by relying on already used principles, while still maintaining one of the biggest advantages of RFID-tags, namely faster checkout. Such technology would be shaped to society, instead of demanding a change of the entire factory-to-customer chain.
 Who invented the barcode?, Barcodes, http://www.barcode.ro/tutorials/barcodes/history.html, accessed on 20-02-2017.
 What is RFID?, Technovelgy, http://www.technovelgy.com/ct/technology-article.asp, accessed on 20-02-2017.
 How RFID works, Technovelgy, http://www.technovelgy.com/ct/Technology-Article.asp?ArtNum=2, accessed on 20-02-2017.
 Contactless Smart Card, Wikipedia, https://en.wikipedia.org/wiki/Contactless_smart_card, accessed on 20-02-2017.
 Monitoring mail with RFID technology, IPC, https://www.ipc.be/en/services/data-solutions/rfid, accessed on 20-02-2017.
 RFID tags, Alibaba.com, https://www.alibaba.com/product-detail/paper-roll-blank-13-56mhz-Mi_60592283328.html, accessed on 20-02-2017.
 Grocery industry operations are facing a real paradigm shift, RFID arena, http://www.rfidarena.com/2013/4/11/grocery-industry-operations-are-facing-a-real-paradigm-shift.aspx, accessed on 20-02-2017.
 Issues of item-level RFID tagging, Quora, https://www.quora.com/Whats-holding-the-grocery-industry-back-from-doing-item-level-tagging, accessed on 20-02-2017.
 Problems with RFID, Technovelgy, http://www.technovelgy.com/ct/Technology-Article.asp?ArtNum=20, accessed on 20-02-2017.
Currently being reviewed by R. Cuijpers.
Currently being reviewed by R. Cuijpers.
Literature study: current systems available
In general, we can state that people don’t like to wait. This means that we should look for ways to reduce waiting time for the users. This will lead to profitable sales and helps return the customer to the stores (Wiseye, n.d.) and thus we will enhance the productivity of an enterprise. We will only focus on reducing waiting times in supermarkets. Therefore, we first should look at what systems are available at the moment, and then we should look for solutions which will reduce this waiting time.
All system services which involve waiting lines of supermarkets can be described in terms of the number of waiting lines and the number of servers (Krajewski et al., 2010). We will discuss each system which is available at the moment in these terms, and then state advantages and disadvantages. In the end, we will draw a conclusion.
Firstly, we have the electronic cash registers and the Point Of Sale-systems. These are systems in which a cashier helps the users and scans the items for them. They will also deal with the payment afterwards. Typical about these kind of systems is that there are multiple lines. This is appropriate when specialized servers are used or when space consideration make a single line inconvenient (Krajewski et al., 2010). In a lot of supermarkets, there is a small amount of space. Thus we could state that the multiple line is convenient for these stores. However, there are also some supermarkets which are gigantic, and still there are multiple lines. This is inconvenient for those stores, since customers will perceive this situation as unfair in terms of equitable waits. That is, the customer can be penalized by picking the slow line (Krajewski et al., 2010). This system has in most cases several server facilities and server proficiency. Thus they have parallel service providers offering the same service. Therefore, these systems are multiserver systems (Krajewski et al., 2010). Obviously, a multiserver system will speed up the whole process. In general, supermarkets will have a multiserver system.
Secondly, we have self-scan systems. There are two versions of these, namely the self-service checkout and the hand-held self-scan. Both these systems are multiserver systems and are also single line systems. This means that the customers will wait in one line, and once it is their turn, they will be divided over the different counters. This system is perceived more fair by the customers, since all customers have equitable waits. Furthermore, the enterprise need fewer cashiers to pay and self-checkout take less space than the single traditional checkout with a cashier.
However, there are disadvantages for both systems. First of all customers experience difficulties with scanning their products when the barcodes don’t scan properly (Jarrett, 2016). Furthermore, customers experience that the amount of space left to pack your groceries is small. Which can make it difficult to put your groceries in a bag after you have scanned them. Lastly, customers feel afraid that they probably have scanned something wrong which makes them feel uncertain and afraid to get arrested by security (Winterman, 2009). For the enterprise these systems are also disadvantageous. Customers can easier commit a theft, since the enterprise lays the responsibility of scanning the products with the customers. There are still two things left which are not seen as a benefit or as a drawback. The first one is the time it takes to scan and pay your products at a self-checkout. With the self-service checkout customers have the feeling that it is going faster since they are doing the work by their selves instead of waiting for a cashier scanning all the products. But in reality it takes you longer at a self-service checkout. The second is that face-to-face time with store personnel can reduce. Since one doesn’t encounter a cashier at the self-service checkout, one will have less contact with store personnel. However, customers don’t experience this as negative, as long as the supermarkets provide a good service (Winterman, 2009).
There are several aspects, we would like to point out for the hand-held self-scan. With this system, the customer scans their products immediately while shopping. This way the customer can keep track of the total cost of their groceries during shopping. The purchaser can place the merchandise in their shopping bag while shopping, which will prevent unpacking at the cashier. The aim of this self-scan is to make shopping faster and easier (Carlberg et al., n.d.). A developer of this scanner is Kroger. His goal was to reduce the wait time at registers at stores. Research has shown that this scanner has reduced the average checkout time in its stores with three minutes and 30 seconds (Anderson, 2013). This can be seen as a benefit for the customers. Besides this reduce of time, customers say that it is easy in use and that the idea of this self-scanning is appealing (Husain, 2016).
Thus in general, we can draw the following conclusions about the self-service checkouts. These systems reduce the waiting times, but not necessarily the check-out times. However, this doesn’t form a problem for the customer, since they don’t experience it this way. Furthermore, these systems are disadvantageous for the enterprise, since the enterprise must lay the responsibility with the customers, which make committing a theft easier for the customer. Thus, these systems do reduce waiting time, but they come with too many disadvantages.
An example of a self-service checkout: 
An example of a hand-held self-scan: 
We should find a system, which will reduce the check-out times and the waiting times, but still lays the responsibility with the enterprise. A solution for this can be an automated check-out system. Thus this comes down to having several scanners which will be able to scan all products by themselves. Lucky for us, such a system has already been invented. Namely, the ‘360 scan portal’ of Wincor Nixdorf. The ‘360 scan portal’ is world’s first automated scanning solution which is practicable in use. It consist of a conveyer belt, the scanner and a customer display. The customer display is used to show the customer how the 360 scan portal works and shows that the customer needs to place the merchandise on the conveyer belt one by one. While the products are on the conveyer belt the barcode scanners scan the product. For this the barcode does not need to be on a certain place, while it can scan the whole surface of the product. This scanner changes the task of the cashier. It cannot exclude the cashier completely. If products have a damaged barcode or a minimum age the cashier is still needed. But it will relieve their time-consuming and physically strenuous task of scanning the products. Which allow them to have more time for the customer friendly service (Wincor Nixdorf, 2011). Wincor Nixdorf has already done a few test runs. Which showed that the 360 scan portal still needed some improvement. Mostly the scan rate needed to improve. The scan rate was already above the 90 percent, but they want to reach a scan rate of 98 to 99 percent (Wincor Nixdorf, 2011).
A 360 degree barcode scanner can have multiple advantages for users and enterprise. The users, customers, will experience benefits with shorter waiting times. Since products can be scanned and processed with an increased speed. The customer will not have to search the location of the barcode on the product. The customer can just place the products on a conveyer belt and the barcode will be scanned automatically. This will lead to more satisfaction for the customer. However, another user, namely the cashiers, will benefit as well. With a cash desk as still used in the shops the cashier has to pick up every product separately which can have medical consequences. While they do not need to scan the products anymore they have more time left to help customer if needed (Vardon, 2015). For enterprise this 360 scanner will be a big investment at first. All current cash desk in the store need to be replaced. But with these new automated scanner, there can be less cashier at the same amount of cash desks. The retailer will have less compensation consequences for the cashiers while they will encounter less medical consequences (Vardon, 2015).
In terms of waiting lines, the enterprise can choose whether they want it to be a single-line system or a multiple-line system. Since the enterprise needs to redecorate their stores to implement these systems, they can also change how they will arrange the waiting lines. Therefore, it can be possible to make a single-line system.
Thus we can conclude that this ‘360 scan portal’ is the perfect system for checkouts. These kind of systems will reduce the waiting times for the users and will result in a better user experience. This is beneficial for the enterprise, since users will return to their supermarkets. Furthermore, the responsibility of scanning the products will still lay with the enterprise, which will make theft less convenient for the customers.
A New Era At The Checkout. (2011, December 09). Retrieved March 02, 2017, from http://www.wincor-nixdorf.com/static/finanzberichte/2010-2011/q4/en/reengineringprocesses/aneweraatthecheckout.html
Anderson, G. (2013, April 25). Why Aren’t More Stores Giving Customers Hand-Held Scanners? – RetailWire. Retrieved March 02, 2017, from http://www.retailwire.com/discussion/why-arent-more-stores-giving-customers-hand-held-scanners/
Carlberg, G., & Karlsson, M. (n.d.). An Evaluation of a Self-Scanning-System in a grocery store environment: improvements, suggestions and further development.
Husain, S. (2016, April 21). AutoID & Data Capture Blog. Retrieved March 02, 2017, from http://www.vdcresearch.com/News-events/autoid-blog/Handheld-Self-Scanning-Solutions.html
Jarrett, C. (2016, December 16). The Pros and Cons of Using Self-Checkouts. Retrieved March 02, 2017, from http://www.businessbee.com/resources/profitability/the-pros-and-cons-of-using-self-checkouts/
Krajewski, L. J., Ritzman, L. P., & Malhotra, M. K. (2010). Operations management: processes and supply chains - Supplement C. Upper Saddle River (NJ): Pearson. Queue management. (n.d.). Retrieved March 02, 2017, from http://www.wiseye.co/?page_id=99
Vardon, J. (2015, August 27). 360 Degree Scanning: Saving Time, Adding Value. Retrieved March 02, 2017, from http://risnews.edgl.com/retail-news/360-Degree-Scanning--Saving-Time,-Adding-Value101858
Winterman, D. (2009, December 09). The problem with self-service checkouts. Retrieved March 02, 2017, from http://news.bbc.co.uk/2/hi/8399963.stm
Literature study: Why hasn't this system been implemented, User perspective
A system which promises faster check outs at supermarkets, which would also mean less time spent waiting in the line at supermarkets sounds like a dream come true for most consumers. So why haven’t the users jumped at the opportunity?
The first reason is for this system to be as fast as possible, the payment methods would have to only be debit/credit card. This is because finding the right money and giving the change all takes away from the speed. Not every user prefers this method of payment and many users may not even own a credit/debit card. This means a percentage of the users is lost immediately.
Users are concerned with waste packaging, which with this new system will become necessary. Reading some user comments on this new technology, it is clear that users are worried about items like cabbage and lettuce getting wrapped simply to make the system work correctly.
Users feel like supermarkets simply want to use this system in order to cut staff members to save money on salaries each month. Frankly this cannot be disproved. People are concerned that the supermarkets are too busy with all the “Techie” stuff and have little to no concern with keeping the prices of the everyday items we need down. This is a great comment, although the waiting times will be decreased if the prices of the items in the store do not decrease or level out. People will shop elsewhere aiming the entire venture a waste of time.
Users feel that the scan while you shop option is a good enough alternative as you have to walk around to collect the items you want in any case. Users think this is faster than having to include the conveyor belt at the end of the system.
Many users want to have the human interaction at the checkout. These users do not just want to be checked out by emotionless robots so that the owners of the stores can make a quick profit by reducing the human staff.
All of the above mentioned points contribute to the reasons why the everyday users hasn’t yet accepted this system. By addressing as many of these points as possible, the user may be tempted into changing their minds. After all there are very few people who enjoy waiting in any kind of line.
Foster, J. (2013, July 18). The super scanner that could kill off queues at the supermarket, and spell the end of 'unexpected items in the bagging area'. Retrieved March 06, 2017, from http://www.dailymail.co.uk/femail/article-2367363/The-super-scanner-kill-queues-supermarket-spell-end-unexpected-items-bagging-area.html
Poulter, S. (2014, May 16). 'Super-till' that could see off traditional checkout worker: New machine is said to be three-times quicker than normal checkout. Retrieved March 06, 2017, from http://www.dailymail.co.uk/news/article-2629916/Super-till-supermarket-staff-New-machine-said-three-times-quicker-normal-checkout.html
Literature study: Why hasn't this system been implemented, Enterprise perspective:
The main aim of any enterprise or venture is to make profits and find solutions to maximize it. These solutions can be an in any form, from cheap labor costs in a foreign country to utilization of technological advancements to reduce the number of paid employees.
An automatic scanning system makes use of present technological developments to automate the complete checkout procedure. In simple words, it provides such businesses a technological substitute to cashiers. This system is just a one-time investment, which can be easily recovered as the need of a cashier at the cash counter will become redundant and the enterprise will not have to pay the cashier anymore. This is how this system benefits an enterprise.
Even though this system increases the profits made by an enterprise, it is still very rarely found in stores/supermarkets across Europe. The reasons why this system was unable to successfully penetrate the European market are:
No system is 100% percent efficient, even though there is no room for human error, the system is still prone to systematic and environmental errors. Environmental errors include exceptional cases where the scanner is unable to scan a product, which can happen because of multiple reasons. So, how to handle such exceptions?
“Wincor Nixdorf has been thinking quite a bit about those questions, a spokesperson said, and its system control software will allow retailers various options for how a no-read will be handled. For example, the system could stop immediately for a no-read to be addressed, or let items pass through until the end, at which point all the no-reads would be handled” .
Handling these exceptions at the end can eventually increase the waiting time as those products need to first be separated from the ones that have been scanned already. Also, since these products were not scanned automatically, they need to be done manually, which in some sense defeats the purpose of the system.
Since there is no cashier at the counters anymore, the possibility of thefts in such stores will rise. Hence, in order to take precautionary measures for such issues the store would need some extra security systems or personnel to perform random checks. This reduces the profit made by the store and also deteriorates the shopping experience of the customers.
Every system in the world is prone to failures, even if it is serviced regularly. This system too can experience breakdowns, and during peak hours such breakdowns can completely shut down the operation of a cash counter. This can cause much longer lines in the rest of the cashier lanes.
Maintenance & Repair
Since the system is operated for about 12 hours a day and is prone to system failures, it needs to be serviced on regular basis. Also, as the system is very technologically complex, the repair work can be very costly and time consuming.
No face to Face Interaction
“We lose the face-to-face with the customer” [ Human cashiers]. Store managers fear that having systems like self-checkout systems reduce the interaction between the customer and the store employees, which according to the store managers is a very significant part of their shopping experience. The automatic scanning system is in this way was very similar to the self-checkout system hence it too faces the issue of no face-to-face interaction.
 Human cashiers still trump self-checkouts for most grocers, Written by: David Sherman.
 Wincor Nixdorf has New Take on Bar Code Scan Tunnel for Grocery by SCDigest.
Literature study: Societal factors
The current situation with respect to point of sale architecture in supermarkets consists of a cashier manning a cash register. The cash register contains a computer, monitor, cash drawer, receipt printer, customer display and a barcode scanner. I.e., all the scanning work is done by the cashier while a computer keeps up with the prices and creates a receipt, which can be printed out and handed to the customer. Therefore, the current situation will always require a cashier .
As a response to the current situation with respect to point of sale architecture, which needs a bulky cash register, a place for the cashier and in most cases, a conveyor belt, smaller point of sale systems have been introduced to the market. This includes, but is not limited to, self-checkout systems in which the customer has to manually select their products for purchase  , self-scanning systems in which the customer has to scan the products while shopping while a system keeps track of all the scanned products to eventually check-out at a cashier  , mobile apps that serve as a checkout system  and hybrid systems, which can be used as a self-checkout system or a traditional system with a cashier .
The main trend that can be seen is that companies are trying to remove the cashier from the checkout by giving the customer more control, providing more convenience for the customer and speeding up the check-out process for people with just a few products. By removing the cashier or at least reducing the amounts of cashiers needed, companies can save money by hiring less cashiers or by hiring them part-time so they are not eligible for all benefits.
The following questions should be answered for an analysis of the societal impact of introducing a new automated EAN scanning system that will reduce the amount of cashiers even more:
• What societal factors are likely to contribute to the implementation of an automated EAN scanning system and what societal factors are likely to impede this implementation?
• What has to change in society to support the transition from the current situation with respect to point of sale architecture to a fully automated one?
Factors contributing or impeding the implementation of a fully automated EAN scanning system
1. Influence on cashiers: number of part-time jobs
As a result of the implementation of semi-automatic systems, the amount of personnel to man a supermarket checkout at any time has decreased. Together with the rise of planning software that takes into account the peak times of customer visits at supermarkets, managers have successfully reduced the amount of personnel at supermarket check-outs at any time. Thus the supermarkets have shifted from hiring cashiers full time, to giving small shifts to more personnel, effectively eliminating full time jobs and substituting these with part-time jobs . Part-time jobs consisting of short shifts are advantageous for store managers because part-time workers do not receive all benefits on top of their salary while a full time worker would and part-time workers generally get paid less per hour . The main disadvantage of part-time jobs is that they require a flexible schedule of the employee: the employee needs to be available whenever the system plans their short shifts. In the current situation, employees often receive their weekly schedule late, meaning they cannot plan any other activities for the next week or they risk losing their shift . Furthermore, declining a shift means that the planning system will schedule the employee even less and thus causing a loss of income and harming the employer-employee relation . This all contributes to a situation in which a lot of part-time employees at a supermarket are struggling to make a living due to few short shifts and low wages. This is disadvantageous from a societal perspective, as this results to people living in poverty while having limited opportunities to improve their situation.
Implementing a fully automated scanning system would reduce the amount of personnel needed even more: no more cashiers manning the self-scanning checkouts would be needed and the traditional cashiers can be removed completely provided that the system can handle any product. However, supervision, especially in the early stages of the implementation, is still needed to prevent theft and to solve system failures. Based on the functioning and success of the automated EAN scanning systems, two scenarios could arise:
1. The scanning system functions properly and there is no need to employ a special team to solve system failures. This would reduce the amount of personnel in a supermarket even further.
2. The scanning system does not function properly and the supermarkets will need to maintain a team to solve system failures as well as employees on stand-by to man the manual check-out stations to compensate for the failure in the automated checkout lines.
It can be concluded that building an unreliable system that has a lot of downtime will only contribute to the necessity of employees with a flexible schedule and therefore increasing the amount of part-time workers, further contributing to the societal disadvantages (i.e. poverty). A reliable system without downtime reduces the total amount of workers and therefore also the amount of part-time workers. This gives the part-time workers the opportunity to look for full time jobs. However, the reason that those employees are currently working part-time at this supermarket is probably that they could not find a better job. This system could also reduce the amount of part-time workers if supermarket managers should decide to hire a few full time supervisors instead of a lot of part-time supervisors.
It follows that implementing a fully automated EAN scanning system does not solve the problem of underpaid part-time workers. However, it does follow that a faulty system will only increase the amount of part-time workers and worsen the societal problems. When designing such a fully automated EAN scanning system, special attention should be paid to robustness of the design to prevent an unreliable system.
2. Number of jobs
In 2014, there were around 3.5 million cashiers employed in the USA . Almost one million of these cashiers were working in a grocery store . This accounts for approximately 0.9% of all the jobs in the USA at that time. Introducing a fully automated EAN scanning system will remove the need for all of these jobs and therefore leave approximately 1% of the population of the USA without a job. Extrapolating this percentage to all first world countries, the number of jobs disappearing is huge. It should also be noted that most personal working in grocery stores is either not able to find another job because of the absence of a degree and is already struggling to make ends meet or is a student that is not able to work another full-time job because of a lack of time. Especially these two demographic groups will have difficulty finding another job. Therefore, introducing a fully automated EAN scanning system will have negative side effects on society through the disappearing of a lot of jobs.
Concerns from society
To investigate the concerns from society with respect to a fully automated EAN scanning system, a parallel with the Amazon Go concept is drawn and this concept will be explored and reactions from society will be summarized, pointing out the important concerns and/or benefits, which can then be linked to the implementation of an automated EAN scanning system.
Amazon Go is an almost fully automated store without lines : by making use of computer vision, information fusion and deep learning, a computer tracks all the products an individual carries around the store. When the customer leaves the store, the computer bills the customer through an Amazon app which is needed to enter the store. There is no need to check-out as the computer keeps track of everything, effectively eliminating waiting lines and the need for cashiers. Thus, in this respect, the Amazon Go store relates to the idea of implementing a fully automated EAN scanning system.
Taking away jobs
The New York post brands Amazon Go as ‘the next major job killer to face Americans’ . Forbes expresses their concerns about the ‘elimination of thousands of jobs’  and even goes as far as comparing the anticipated public unrest to the French Revolution. According to Britt Beamer, president of America's Research Group, ‘about 75% of typical grocery store staff may stand to lose their employment’ . Some nuisance is added by saying that Amazon Go will likely create as many jobs as traditional grocery stores according to ‘people familiar with the matter’ . However, most of the sources are certain Amazon Go will eliminate the need for a lot of jobs.
In this aspect, an EAN scanning system has the same effect as Amazon Go: eliminating a lot of jobs. However, automation has always been a cause of the elimination of jobs, especially those only requiring little education . In turn, automation provided other jobs . Therefore it can be argued that the problem of job elimination is not a problem that can be solved by automation itself and thus cannot be solved by designing an EAN scanning system that somehow creates jobs in a grocery store, as this defeats the whole purpose of such a system. The problem of job elimination can, according to  and , be solved by better education for everyone and retraining workers for a new labour market.
This is implying that the current education system has to change in order to implement automation that will put a lot of jobs on the line.
Another concern voiced about the Amazon Go system is one with respect to privacy: customers are being watched continuously during their shopping time by a computer system in order to register the products they carry. Although this is a very interesting topic, it is not applicable to an automated EAN scanning system, as this system does not require tracking customers while they are shopping.
Other factors that influence the utility of society
• Cashiers working in a sitting position at a traditional point-of-sale architecture are often experiencing musculoskeletal disorders and discomfort . By introducing an automated EAN scanning system, these kinds of injuries will be prevented.
• Having social interaction with a cashier while checking out has a more positive affect on happiness than being as efficient as possible . This indicates that customer interaction in stores is an important element of customer satisfaction.
 Point of sale, https://en.wikipedia.org/wiki/Point_of_sale#cite_note-25, accessed on 3-3-2017
 Self-checkout, https://en.wikipedia.org/wiki/Self-checkout, accessed on 3-3-2017
 Supermarket self-checkout kiosk, http://web.mit.edu/2.744/www/Project/Assignments/humanUse/lynette/2-About%20the%20machine.html, accessed on 3-3-2017
 Self-scan gadgets, https://www.wsj.com/articles/SB10001424052748703421204576329253050637400, accessed on 3-3-2017
 Mobile point of sale apps, http://www.scandit.com/mobile-point-of-sale-apps-redefining-the-industry/, accessed on 3-3-2017
 A Part-Time Life, as Hours Shrink and Shift, http://www.nytimes.com/2012/10/28/business/a-part-time-life-as-hours-shrink-and-shift-for-american-workers.html, accessed on 3-3-2017
 Effects of sitting versus standing and scanner type on cashiers K. R. LEHMAN² , J. P. PSIHOGIOS² * and R. G. J. MEULENBROEK³ ² NCR Corporation, Duluth, GA 30096, USA ³ Nijmegen Institute for Cognition and Information (NICI), University of Nijmegen, PO Box 9104, 6500 HE Nijmegen, The Netherlands.
 Is Efficiency Overrated? Minimal Social Interactions Lead to Belonging and Positive Affect, http://journals.sagepub.com/doi/abs/10.1177/1948550613502990, accessed on 5-3-2017
 Bureau of Labor Statistics, U.S. Department of Labor, Occupational Outlook Handbook, 2016-17 Edition, Cashiers, https://www.bls.gov/ooh/sales/cashiers.htm, accessed on 5-3-2017
 Occupational employment statistics, Bureau of Labor Statistics, https://www.bls.gov/oes/current/oes412011.htm, accessed on 5-3-2017
 Amazon Go, https://www.amazon.com/b?node=16008589011, accessed on 6-3-2017
 Amazon introduces next major job killer to face Americans, http://nypost.com/2016/12/05/amazon-introduces-next-major-job-killer-to-face-americans/, accessed on 6-3-2017
 Amazon's Grocery Would Eliminate Thousands Of Jobs, https://www.forbes.com/sites/eriksherman/2016/12/07/amazons-grocery-would-eliminate-thousands-of-jobs/#6c4e1f3a2477, accessed on 6-3-2017
 Amazon Go Stores Facing False Job Killing Charges?, https://www.benzinga.com/news/16/12/8791565/amazon-go-stores-facing-false-job-killing-charges, accessed on 6-3-2017
 Amazon's new grocery store highlights a huge hole in Donald Trump's promise on jobs, http://www.businessinsider.in/Amazons-new-grocery-store-highlights-a-huge-hole-in-Donald-Trumps-promise-on-jobs/articleshow/55823424.cms, accessed on 6-3-2017
 Google chief economist: Don't fear the robots, http://www.businessinsider.com/google-chief-economist-dont-fear-the-robots-2015-2?international=true&r=US&IR=T, accessed on 6-3-2017
 Report Suggests Nearly Half of U.S. Jobs Are Vulnerable to Computerization, https://www.technologyreview.com/s/519241/report-suggests-nearly-half-of-us-jobs-are-vulnerable-to-computerization/, accessed on 6-3-2017
When developing a new product, it is important to thinks about certain important design aspects. These are the so-called RPC’s: Requirements, Preferences and constraints. This section describes the RPC’s of the final product that could be used in the supermarket.
The requirements of the APS system are the following. The APS system should not increase the average time needed for a single scan, therefore scanning must be possible in under 3 seconds. A second important requirement is that the system should not increase the overall cost for a supermarket. To determine this, attention has to be payed to the overall effect such system has on a supermarket. For example if the system reduces the personnel costs of a supermarket but the waiting times are significantly increased, customers could decide to go to another store, therefore decreasing the turnover and possibly the profit of a supermarket. The same reasoning applies to a super-expensive system that has very fast scanning times. More customers could decide to go to a store with such a fast checkout system, increasing turnover of the store. An important factor on the profit or loss by using the APS system is the amount of unscanned products. The more unscanned products, the higher the loss for a supermarket. However, a ridiculously low unscanned to scanned product ratio will most probably require long scanning times, conflicting the first requirement. It is clear that a good trade-off should be made. Thirdly, the system needs to be safe for customers and supermarket personnel, and should not be able to hurt human beings in any way. The sound produced by the system cannot be more than general noise levels in supermarkets (~80dB). Note that especially the second requirement is hard to investigate, and could be different for different supermarkets. It is beyond the scope of this project to determine the exact maximum price of the APS system.
Furthermore there are some preferences. First of all: scanning time should be as short as possible, for obvious reasons. Secondly the system should be as cheap as possible. The system should also increase the profit of supermarkets as much as possible. Thirdly, unscanned to scanned product ratio of the system should be as small as possible. Fourthly, the system should require minimal changes to the currently used checkout system. The technology behind it should be as cheap and as simple as possible, to reduce possible repair costs. The system should be reliable and should have down-times that are minimized. A visually appealing design is preferred. The system should make as little noise as possible. Maintenance should be easy to perform. The system should be as energy efficient as possible. Lastly, there are a constraint to the APS system. The system must be installable in all major supermarket chains in terms of size, power requirements etc. without having to change the current infrastructure. The cost of the system could be a constraint in the case that a specific customer has asked to design and produce the APS-system. Of course, this list is not exhaustive. It could be extended by questioning more supermarkets about their opinion on such a system and possible design flaws of currently used systems, we are unaware of.
However, due to time and budget constraints, only a prototype can be made instead of a finished product. This prototype also has a number of RPC’s, which are given below. Note that the prototype will be tested in simplified conditions. This is because it is only meant as a proof of concept, whose design can be elaborated, instead of a ready-to-use solution. Therefore, the RPC’s are focused on the limited testing conditions. Test will only be done on a limited range of products. It is assumed that the barcodes are in any orientation or position, except facing down. Dropping these assumptions will lead to the need for more expensive hardware and significantly more time, which is unfortunately not possible in this project.
The prototype is required to be able to identify a limited selection of products. The prototype needs be built with materials supplied by the university (Arduino’s, servos, DC-motors, small electronics etc.) and self-bought materials in a limited budget. The prototype must be designable and realizable in 5 weeks. Preferences are: as cheap as possible, as fast as possible, nicely finished and lowest possible unscanned – scanned product ratio. Furthermore, it would be nice if an exhaustive database of barcodes could be used to identify products. A design which can deal with barcodes that are facing down is preferred over one that cannot do so. Lastly, it would be nice if a conveyor belt could be used to transport products through the scanner, in order to simulate real-life situations as good as possible. The only constraint regarding the prototype is the price of self-bought materials. A maximum price of €100 is agreed on. The prototype will be developed, bearing these aspects in mind.
Questionnaire for the Enterprise
Product Description: The product is called Automatic Product Scanner (APS). This product is based on the fact that all the items in a store nowadays have a barcode on them which can be used to identify the model number and the cost of the item. In order to get this information from the barcode, a barcode scanner can be used. APS consists of three such scanners, which continuously keep moving and have the ability to scan 70 times per second. This process is completely automated via computer. So products are placed on a conveyor and the customer is given a receipt, all done by a computer. This would enable shorter queuing times.
Q1) Do you think a system like this can reduce the need for a cashier ?
a) Definitely yes.
b) Yes, I think so.
c) No, that cannot happen.
d) Maybe, I am not sure.
Q2) Do you think a system like this can help you reduce the long lines at cash counters during the peak hours of the store ?
Q3) Are there any add-ons you would like to have on this system ?
b) Yes, namely:
Q4) By how many minutes would this technology need to reduce the average waiting time for a customer in your stores for you to consider employing it??
Q5) How many regular checkout lines do you have at the moment? How many self-scan?
Q6) Do you think that use of such systems will improve the shopping experience of your customers ?
a) Yes, surely.
b) I think yes.
c) Maybe not.
d) Definitely not.
Q7)Do you agree that such systems can help you maximize the turnover of your store?
a) Yes, surely.
b) I think yes.
c) Maybe not.
d) Definitely not.
Q8) Would you be willing to purchase this system for your stores ?
a) Yes, I would.
b) No, I would never buy this system.
c) I am not sure.
Q9) How much are you willing to pay to employ this system in your stores ?
Q10) How many such machines would you like to have in your store ?
a) Between 1 and 3.
b) Between 4 and 6.
c) More than 6.
Questionnaires for the Users
Waiting times in super markets (is frustrating)
Waiting in supermarket checkout lines is an inconvenience. This questionnaire will be regarding the waiting times experienced in supermarket checkout lines. There will be five questions each based on a scale from one to five, one being highly disagree and five being highly agree. Please fill in the number next to the question.
1. I spend a long time in checkout lines on average.
2. My shopping experience is spoiled due to long waiting times.
3. I avoid busy times in the stores.
4. I am annoyed about how long I have to wait in the checkout line.
5. I would shop at a different store simply to avoid the queueing times
The APS System
The APS or Automatic Product Scanning system will be used to decrease the checkout time at supermarkets by a substantial amount. There will be five questions each based on a scale from one to five, one being highly disagree and five being highly agree. Please fill in the number next to the question.
1. I would like that waiting times in checkout lines decreased.
2. I feel that interaction with an actual person is important during checkout.
3. I would enjoy using this system at the checkout in my local store.
4. I feel this system would be a success at my local store.
5. My shopping experience will be improved by implementing the new system.
To make a simulation of the supermarket, the language Chi 3.0 was used. This language runs on Eclipse. The whole software package is available at .
Point of sale architecture overview and explanation
Note: A schematic overview of the point of sale architecture in supermarkets can be found at a shared dropbox folder with all the files for this project, under the name PointOfSaleStructure.pdf.
The point of sale architecture describes the process of customers arriving with products (in a cart or a basket) at a checkout. The customers can choose between check-out points manned by a cashier and automated check-out points by means of an automated EAN scanning system. When there is a queue, customers have to wait in line before they are served at either station. After the line has disappeared, the customers use both of their hands to put their products on a conveyor, i.e. two products per lift. The conveyor transports the products to either the cashier or the automated EAN scanning system, where the products are scanned one by one. Occasional failures occur, both at the cashier (by failing to find the barcode sometimes) and the automated EAN scanning systems (by downtime or scanning problems). After a product is scanned, it is transported by a conveyor belt to a point where the customer can pick up their products. The customer gets a receipt from the cashier or indicates on the user interface of the automated EAN scanning system that he/she would like to pay, pays and leaves the supermarket.
Building blocks and assumptions of model (in progress, might change as model proceeds)
Generates customers based on inter arrival times of customers.
• Assigns a day and time entered to each customer based on the current time of the model.
• Assigns number of products of a customer based on two normal distributions: one for big groceries with a mean of 50 products and a standard deviation of 10 and one for small groceries with a mean of 10 products and a standard deviation of 3. The choice between the two distributions is made by flipping a fair coin (i.e. 0.5 for small, 0.5 for big).
• Assigns an age to the customer based on a normal distribution with a mean of 35 and a standard deviation of 20.
• Assigns the inter arrival time of customers based on the current time of the day: customers are assumed to arrive according to a Poisson process (meaning memoryless) with a mean which is dependent on the current time of the day. During peak hours (10:00-12:00, 13:00-15:00 and 18:00-20:00), customers arrive with a mean inter arrival time of 1.0 minutes. During off-peak hours (8:00-10:00, 12:00-13:00, 15:00-18:00), customers arrive with a mean inter arrival time of 3.0 minutes.
The conveyor transports the products placed onto it by the customer to the cashier. The conveyor belt has a length of 1 m and products move with a constant speed 0.3 m/s. It has a maximum capacity of 30 products
A cashier takes a product from the conveyor belt, scans it with a time according to a normal distribution with a mean of 4.0 seconds and a standard deviation of 0.5. After that, the product is sent to another conveyor belt. Further things that could be modelled in the cashier step are:
• Once every e.g. 100 products, the cashier fails to find the barcode and takes 10 seconds instead of 4.
• Payment interactions with the customer.
Automated EAN scanning system
An automated EAN scanning system functions according to the same basic principles as a cashier: it takes a product from the conveyor, scans it with a time according to a normal distribution with a mean of 2.0 seconds and a standard deviation of 0.25 (values will follow from prototype). After that, the product is sent to another conveyor belt. Further things that could be modelled in this step are:
• Once every e.g. 100 products (follows from prototype), the automated EAN scanning system cannot correctly scan the barcode and supermarket staff has to interfere, causing extra delay.
• The automated EAN scanning system is subject to random downtimes caused by electronic or mechanical malfunctioning, causing it to be out of business for a few hours.
The exits process is the last process in the chain and calculates and prints all relevant parameters: mean flowtime of customers, amount of customers in the store and throughput of customers. It can also print individual data of customers such as age, number of products and time spent in the store.
The parts that are needed for the prototypes have either been bought or ordered. The wooden body panels are currently being lasercut by M. van Gorp with the delivered CAD-drawings. The axis on which the gearn will be mounted have been ordered, together with the nuts that will be needed. Wood glue has been bought. Pictures of the CAD-drawings of the body panels, can be found under the following links: , , , , 
Interview questions: The self checkout system
1. Do you like the idea of self checkout areas in the supermarket?
2. Which checkout option do you prefer: the cashier or the self checkout?
3. Do you think that human interaction during checkout is important?
4. Do you think that the self checkout system will eventually replace the human cashier and why?
5. Would you shop at a store which only has the option for a self checkout?
Security/Anti-theft measures in Supermarkets:
It has already been mentioned several times that implementation of automatic scanning systems in supermarkets can lead to an increase in thefts as the introduction of such system reduces the level of security in supermarkets.
The autonomous scanning system in many ways is quite similar to the self-checkout system already implemented in a large number of stores, hence the ploys of stealing that are used for self-checkout systems such as:
a) Leaving expensive items in the shopping cart, or in the reusable bag on the floor, while scanning cheaper ones [How to stop];
b) Scanning items with the bar code facing up, or covered by your hand [How to stop],
c) Scanning an item with one hand while dropping another item into a reusable bag on the floor [How to stop]
can also be used for such systems.
In order to prevent this from happening, a store chain and system designers can take multiple precautionary measures such as the ones discussed below.
The firm designing the scanning system can implement some weight sensors in the system, which can be used to detect un-scanned products. It could be done in the following way; Before putting the products on the conveyor belt, the baggage containing all the items is first weighed and then, when the products are kept on the conveyor their weight is measured individually by weight sensors included in the system which is then summed up at the end and the total weights before scanning and after scanning can help detect any missing products. Having such a system can sometimes affect the shopping experience of the consumers due to false alarms.
Also, CCTV cameras can be implemented at each cash station such that every customer can be monitored during the check-out procedure. Besides that, by making use of video analytics one can easily detect any signs of theft and sweethearting (the process of letting folks walk away with stuff with a wink from the cashier) [Stoplift watches].
Another possible way to avoid thefts is random security checks by some of the staff members of the store and keeping high fines or punishments for thefts, so that the thieves are aware of the fact that there is a possibility they can be confronted anytime randomly. However, this sometimes completely ruins the shopping experience of non-stealing customers.
 How to Stop Those Self-Checkout Thieves, Written by Malay Kundu, StopLift Checkout Visions Systems
 Stoplift watches you at the Self-Checkout, by John Biggs.
Program for APS system:
We started working on a program for the APS system. This program will have 2 main components. One component is to take the scanned barcodes as input, search the right product for it, and put it on a list of bought products. The other component will be the graphical user interface which will be shown to the customers of the supermarkets. The former can be formed wihtout the results of the questionairres, since the backend won't be visible. For the frontend we have to take the opinion of the users into account.
Our plan is to focus first on getting the program to work. Therefore it won't be necessarily userfriendly, but it will work. Secondly we will take the output of the surveys into account and adapt the program to the needs of the user. We know that this is not a user-centered design, which has our preference. But since the surveys are delayed, we don't have input from the user yet, and we cannot adapt our program to the needs of the user.
We have created a github repository. This makes it easier for us to collaborate on coding, and it will also other people to view our progress. Our github repository link is https://github.com/Ava-S/USE-aps-system.
All parts for the prototype have been processed. The parts have been assembled and the conveyor belt has been made. The frame for the scanners is also made and it is adjustable in width, such that we can alter the width for optimal functioning of the prototype. Photos can be shown during the discussion hour (if wanted), if not, it can be seen during the final presentation. The conveyor will not be brought to the discussion moment, because of its size.
The following alterations were made with respect to the simulation described in week 4:
• A possibility has been included to model that the automated EAN scanning system cannot correctly scan the barcode and supermarket staff has to interfere, causing extra delay. This has been modelled as a chance experiment with a chance on failure of 0.1%.
• Payment interactions are included in the cashier and APS system, i.e. it takes time for the customer to pay after all the products have been scanned.
• A conveyor has been added that receives products and transports them to the following process. The transport time is set as a timer in the process. I.e., the first conveyor takes products from unloading customers and sends them to the cashier. The second conveyor takes products from the cashier and sends them to the buffer at which the customers bag their products.
• An unloading process was made: this process consists of customers putting their products on a conveyor. After putting all products on the conveyor, customers are sent to the cashier. The unloading process asks for customers from the general buffer, sends customers to the cashier and sends products to the conveyor.
• A general buffer was made that accepts all customers from the generator and then sends them to the first unloading process asking for a customer, i.e. always the shortest waiting times.
• Payment interactions with the automated EAN scanning system and cashiers. This basically means there is some time before the customer can go and bag his products and effectively delaying the next customer unloading his groceries.
• A process was made that effectively creates one check-out system, i.e. an unloading process, a conveyor process, a cashier or APS, a conveyor and a loading process. This process will be used in model building.
• The manager process will not be modelled, as this will be replaced by different experiments of having a different amounts of cashiers/APS systems.
What remains to be done:
• Set up multiple experiments to get outputs for different setups of cashiers and APS systems so that data can be compared.
The results for the simulation were derived under the following settings:
• Age is distributed according to a normal distribution with mean 35 and a SD of 15. Currently, age is not used in the model. It could be used as an indication for the tech-savviness of people as an improvement to the model.
• Small groceries have an estimated mean of 10 products with a SD of 3.
• Big groceries have an estimated mean of 50 products with a SD of 10.
• 2/3 of the customers shop for small groceries, while 1/3 shop for big groceries. This was estimated based on shopping behavior of relatives: once a week for big groceries and twice extra for some small groceries.
• The store is opened between 8 AM and 8 PM.
• Peak times are between 10:00-12:00, 13:00-15:00, 18:00-20:00.
• Down times are between 8:00-10:00, 12:00-13:00, 15:00-18:00.
• During down times, the time between two customers is exponentially distributed with a mean of 2 minutes to model memoryless and independent characteristic of the arrival rates.
• During peak times, the time between two customers is exponentially distributed with a mean of 20 seconds.
• It takes a customer 1 second to unload one product onto the conveyor. This could be expanded upon by allowing the customer the possibility to unload 2 products at the same time and by including a SD for the unloading time.
• The cashier scans products with a time that is normally distributed with a mean of 5 seconds and a SD of 0.5 seconds. The time to pay a cashier is 10 seconds. This could be expanded upon to add a chance that the customer cannot find his money and therefore payment takes longer.
• The APS scans products with a time that is normally distributed with a mean of 3 seconds and a SD of 0.5 seconds. The time to pay at an APS is 20 seconds. The APS also has a chance of 0.0001 per product of randomly breaking down. When the APS breaks down, maintenance has to be carried out. Maintenance time is normally distributed with a mean of 10 minutes and a SD of 4 minutes. This process could be expanded upon by adding more problems customers could have with the APS.
• The conveyor takes 10 seconds to transport all products. The maximum capacity of the conveyor is 60 products. An improvement in this process could be to have a variable conveyor time dependent on the place at which the products are placed, which is a function of the current amount of products on the conveyor.
• The buffer at the end of the second conveyor after the cashier/APS has a maximum capacity of 100 products. This buffer size does not have a lot of influence, as picking up products will always be faster than scanning products at the cashier.
• It takes a customer 1 second to load one product into his bag from the buffer.
Note that these conditions can be altered to better suit the specific statistics of supermarkets. These conditions are mainly estimates because there was no information on handling times on the internet. The code for the model can be found at .
The model has been run for a range of 1 to 10 check-out points. At each check-out point iteration, all possible combinations of amounts of cashiers vs amount of APS were simulated, resulting in a total of 65 (sum i=1 to i = 10 (i+1)) data points. For the sake of convergence of the mean flowtimes, throughputs and wip, 20000 customers were simulated per combination. Mean flowtime and throughput of customers with small and big groceries were calculated separately as well as the mean flowtime and throughput of all customers, including the average WIP (work in progress) in the system. The results are shown in the Excel file DataSimulatie.xlsx, which can be found on .
A first analysis on the raw data is done, containing some general trends and a little discussion. Figures to analyze the data are currently being constructed but are not yet available.
The simulation predicts lower flowtimes and WIP for systems which have an increasing percentage of APS for all numbers of total check-outs. The effect is more noticeable in some setups than in other because of the amount of total check-outs. I.e. having 10 cashiers is easily enough to handle all incoming customers, so replacing a cashier with an APS will have a tiny influence. However, when the amount of cashiers is not enough to handle every incoming customer at once, adding an APS will reduce the flowtime of customers and decrease the WIP, which is beneficial for waiting times and customer satisfaction. Other options are increasing cashier speeds or reducing the variability of the process times.
The throughput of the system is not heavily influenced by the amount of cashiers vs the amount of APS except for the case of one check-out in total. This is because mean throughput in a steady state system (i.e. no overflowing of general demand buffer) is determined by the inflow of customers, which is fixed. All deviations at 2 or higher check-outs in total can be attributed to chance because inflow is generated with certain probabilities. The case of one check-out in total is an unstable system, causing the demand buffer to overflow. In this case, the process rate of the cashier is the limiting factor to the throughput of the system.
Still a significant amount of time in the supermarket (both for APS and cashier) are spend waiting. The mean flowtime for small groceries for 10 APS and a total of 10 check-outs is 229.4 seconds, a little less than 4 minutes. In the extreme case of having 20 products for small groceries, the process time alone is estimated at 190 seconds as a maximum: 20*1 (unloading) + 10 (conveyor) + 6 * 20 (scanning, overestimated) + 10 * 2 (conveyor, all products but last two have already arrived at buffer) + 20*1 (loading) = 190 seconds. This leaves an average lower bound of 17% of the total time as waiting time. However, doing analogous calculations for big groceries at 10 APS, leaves the mean flowtime of big groceries almost equal to the total flowtime (this, of course, depends on the specific amount of products). As an improvement, different lines in the supermarket could be introduced for big and small groceries. This advantage is already being used by the self-checkout and self-scan systems (since self-scan does, in most of the cases, not require any scanning at all at the check-out)
More detailed analysis, conclusions and a discussion will follow.
Goals of this project
Since the goals of our project appearantly were not clear, they get repeated here:
The goal of this project is to investigate the feasibility of implementing automated International Article Number (EAN) scanning systems in supermarkets. The objectives of this project are (1) to give a well-founded conclusion on the feasibility of a fully automated EAN scanning system and (2) to deliver a working prototype as a proof of concept. In order to successfully complete this research assignment, the following goals have been drawn up. At each goal, it has been indicated how this contributes to achieving the end goal. • Conduct a user study into the demand of automated EAN scanning systems in supermarkets. This is necessary to determine possible shortcomings and user concerns with respect to the new system. These concerns can be addressed in the prototype.
• Conduct a study into the interest of enterprise in such automated EAN systems. With the insights gained from this study, the prototype can be tweaked to overcome concerns from enterprise.
• Conduct a study into the advantages, possibilities and limitations of automated EAN scanning systems with respect to waiting time reduction, throughput and the effect reliability of the system has on these parameters.
• Design and produce a small scale prototype/proof of concept of an automated EAN scanning system to demonstrate the operating principles of such a system. During the making and testing of this prototype, shortcomings can be detected and addressed in a recommendation on a possible next prototype.
In order to reach these goals, the following sub goals have been set:
• Conduct literature research into the development and application of already existing automated EAN scanning. Also investigate possible competing technologies.
• Design a questionnaire to measure the extent of frustration currently experienced by the user while shopping in supermarkets and especially while waiting in line. Also measure the extent of utility the users would derive from having to wait in line shorter as a result of the automated EAN scanning system.
• Conduct literature research into the business model currently employed by big supermarket chains with respect to:
o Cashiers: payment, efficiency, advantages, problems and limitations.
o Current business model: Limitations, possibilities to improve and reasons why it evolved to be this way.
o Transition to new system: what restrictions does the current business model set on the implementation automated EAN scanning systems.
• Visit local supermarkets to do research into their opinions and ideas of using an automated EAN scanning system.
• Search data about waiting time and consumer behavior in supermarkets and conduct a statistical analysis of the data:
o Waiting time
o Products bought
o Average visits per week
• Do a study on the implications and effects of introducing an automated EAN scanning system into supermarkets.
• Model a supermarket with and without an automated EAN scanning system by making use of the involved statistics. This model can then be used as a guideline to test the influence of implementing an automated EAN scanning system on waiting times.
The results of this research assignment will be: • A summary of the conducted user and enterprise study into the demand of users and enterprises for the automated scanning systems.
• A summary of the study done into the user and enterprise aspects of implementing an automated EAN scanning system in supermarkets.
• A short report on the technical possibilities of the automated scanning system.
• A short overview of possible competing technologies.
• Demonstration of the concept with a simplified prototype.
• A model predicting the effect of the concept on waiting time.
The prototype is developed, bearing the RPC's in mind (see earlier). A video of the realized prototype will be shown durig the discussion moment. It is entirely made out of wooden plates that are laser cut on our specifications. The low-cost scanners have been attached to the frame. Furthermore, a conveyor belt has been made with again laser cut wood, metal rod and caterpillar tracks. The reason to do this was to simulate real-life conditions more accurately and t give a clear demonstration of the concept. Users and enterprises could be confronted with the entire system. Their reactions to the system could then again provide useful input for a next design. Note that due to time constraints, this step has unfortunately not been possible to implement in the design cycle.
User questionnaire analysis
Pie charts of the responses were created to see an overview of the results of each question.
APS questionnaire analysis:
• 101 people responded to this questionnaire.
• The oldest person who answered was 66 years old.
• The youngest person to answer was 17 years old.
• The average age of people who answered was 25 years old.
• The age with the most responses is 20 years old with 21 responses.
Waiting in line questionnaire analysis:
• 61 people responded to this questionnaire.
• The oldest person who answered was 61 years old.
• The youngest person to answer was 17 years old.
• The average age of people who answered was 26 years old.
• The age with the most responses is 20 years old with 10 responses.
Improvements suggested by the public:
• Make sure the security is good.
• Think about the affect it will have on the current staff in terms of job satisfaction and future position.
• Check out Amazon Go.
• The interface must explain the procedure well otherwise the lines will increase instead of decrease.
• The older generation will not welcome the system as the younger generation do.
The link to the pie charts file can be found here: 
The public does not enjoy waiting in the long lines at supermarkets, with many going as far as to say they are annoyed by how long they have to wait. Many consumers will go out of their way to avoid prime time at the shops to avoid lines. Finally peoples overall shopping experience is spoiled by having to wait long at the checkout. Regarding the opinion of the new system, many people are optimistic about it and feel it could be a success in their own stores. The majority of people do not care about human interaction and feel that they would enjoy using the system themselves. Finally a large portion of people feel that in fact this system would improve their overall experience. It is clear that the majority of people who responded to these questionnaires are in their twenties and early thirties. Although there were a few in their late fifties and early sixties, most of the people are of a younger generation. This means that the younger generation are feeling optimistic about the new technology which could potentially be brought to supermarkets. This is useful information as if the current generation can learn to use it then the new generation will have no problem using the system as they will have grown up with it. The responses that were received are all pointing to the fact that the public would be willing to test out the system, even if it means reducing the amount of human contact during checkout. The people are willing to back the idea as long as some of the suggestions given and reasons for failure of previous versions are taken into account. The most important of these suggestions is to have extremely clear and friendly instructions, make sure that the security is tight and finally ensure that current employees are not out of a job because of the system.
This document  is our final report.
Have fun while reading!