PRE2020 1 Group1

Group members

Name	Student ID	Department
Davide Fabbro	1255401	Electrical Engineering
Lieke Nijhuis	0943276	Innovation Sciences
Ikira Wortel	1334336	Electrical Engineering
Wout Opdam	1241084	Mechanical Engineering

Problem statement and objectives

COVID-19 is a disease that as of 5 september 2020 has 26.171.112 confirmed cases and 865.154 confirmed deaths worldwide. This pandemic has caused many countries to go into lockdown, and strict measurements are being taken to lessen the spread of the disease. The virus easily spreads itself in crowds, and this is where the idea of a COVID-19 symptom detection system in choke points and crowded places was born. In this study we will focus on the Heuvel Galerie Eindhoven in particular.

The objectives of this center are:

- Detection of COVID-19 symptoms in a crowd

- Suggesting screened people to take a conventional test (swab test) and possibly denying them access

- Applying a form of social control in order to make people follow the rules better

- Creating a general feeling of awareness and safety

The project

This project is part of the course 0LAUK0 Robots Everywhere. The project concerns a COVID-19 symptom detection system for choke points and crowded places.

Approach

A literature study will be performed on the feasability of a COVID-19 symptom detection system for choke points and crowded places. This literature study will include an investigation of possible privacy issues that can arise when such a system is deployed in public places.

A design for the hardware of the system will be made and a pseudocode will be written for the software behind the system. The system will be designed as to avoid the possible identified privacy issues.

Also, a cost-benefit analysis will be performed to get a view on the financial feasibility of the system.

Deliverables

The deliverables for this project include:

• Literature/feasability study of a COVID-19 symptom detection system for choke points and crowded places, including a privacy-issue study.
• Design of the hardware of the system, including posters.
• Map of the Heuvel Galerie with installment locations.
• Simulation of the effect of the system on a flow of people.
• Pseudocode of the software of the system.
• Cost-benefit analysis.
• A wiki page.
• A presentation video.

Planning and milestones

In the figure below, the project planning is given. The left most column contains the milestones for the project. The column to the right of the first column contains all tasks that need to be completed in order to achieve that milestone.

The planning for the project.

Here, an overview will be given of the tasks that need to be completed per week (corresponding to the figure above), and who is responsible for them:

Week 1

Task	Person	Completed
Brainstorm about subject	Everyone	Yes
Create layout for wiki page	Wout	Yes
Make planning for project	Wout	Yes
Write problem statement and objectives	Lieke	Yes
Describe users	Ikira	Yes
Describe user needs	Wout	Yes
Read and summarize articles for State of the Art	Everyone	Yes
Write State of the Art	Davide	Yes

Week 2

Task	Person	Completed
Describe psychological effect of visible detection system	Lieke	Yes
Create layout for entire wiki page	Wout	Yes
Update planning of project	Wout	Yes
Redefine USE'ers	Ikira	Yew
Redefine USE'ers needs	Wout	Yes
Describe possible use case (Heuvel Galerie)	Wout	Yes
Research possible privacy issues and describe on wiki	Ikira
Describe what happens after test (deny acces?, regulations?)	Lieke	Yes
Describe different detection options, what we chose and why, why no 1.5m detection	Davide, Lieke
Re-write (in wiki) project focuss only covid symptoms	Davide	Yes
Read/investigate similar systems and describe on wiki	Davide	Only Thrm. Img

Week 3

Task	Person	Completed
Redefine users needs based on questionaire and formulate RPC's	Everyone
Make sketches of detection system	Wout
Make sketches of measurement station	Wout
Checkpoint description (research conventional checkpoints) and sensors placement	Davide
Research possible privacy issues and describe on wiki	Ikira
Describe psychological effect of visible detection system	Lieke
Questionnaire for mall- and shop owners of Heuvel Galerie	Lieke	Yes
Questionnaire for shopping people in Heuvel Galerie	Ikira, Lieke	Yes
Design posters	Lieke	Yes
Study on coughing movement	Davide
Read/investigate similar systems and describe on wiki	Davide
Matlab simulation for throughput of system	Ikira
Netlogo simulation for effect on people flow	Wout
Research legal side of health checks at entrance	Wout	Yes

Week 4

Task	Person	Completed
Sensors placement and type of microphone	Davide
Perform cost-benefit analysis and describe on wiki	Lieke
Write pseudocode for notifying medical staff of person with symptoms	Ikira
Design posters	Lieke
Discuss required software functionalities	Everyone
Write pseudocode for detection system	Ikira
Matlab simulation for throughput of system	Ikira
Netlogo simulation for effect on people flow	Wout
Make sketches of detection system	Wout
Make sketches of measurement station	Wout

Week 5

Task	Person	Completed
Perform cost-benefit analysis and describe on wiki	Lieke
Conclude on feasability and describe on wiki	Ikira
Make CAD design of detection system	Wout
Make CAD design of measurement station	Wout
Write pseudocode for measurement station	Ikira
Discuss and finalize sketches	Everyone, Wout

Week 6

Task	Person	Completed
Conclude on feasability and describe on wiki	Ikira
Describe CAD designs on wiki	Wout
Put pseudocodes on wiki with description and user interface screenshot examples	Ikira
Discuss what should be in video	Everyone
Make a script for video	Lieke
Make slides and visuals for video	Lieke
Write conclusion	Davide
Write discussion
Write recommendation
Check wiki completeness (titel, introduction etc.)	Everyone
Make CAD design of detection system	Wout
Make CAD design of measurement station	Wout

Week 7

Task	Person	Completed
Record video	Lieke
Describe peer review	Ikira
Finalize layout of wiki	Everyone
Final check: spelling, citation style etc.	Everyone

Week 8

Task	Person	Completed
Hand everything in	Everyone

Users

In this section the use case for the detection system will be given and the users and their needs are identified.

Use case: Heuvel Galerie

In the projected use case for the COVID-19 symptom detection system, the system is deployed at the Heuvel Galerie (official new name: Heuvel Eindhoven). The Heuvel Galerie is a big indoor shoppingcenter located in the center of Eindhoven.

Heuvel Galerie, western entrance .

The Heuvel Galerie has three floors, four entrances and it's own underground parking facility. The Heuvel Galerie is a big shoppingcenter: it has over 80 stores and houses the Muziekgebouw Eindhoven and a Holland Casino. Because it houses many stores and facilities, it draws a significant number of visitors. In 2001 the Heuvel Galerie got an average of 200.000 weekly visitors . Because of the corona situation, the shoppingcenter has taken measurements like putting up hand-disinfection stations and creating walking routes. This enables the shoppingcenter to remain open and draw visitors. The following table (in Dutch) shows a recent (as of 7-9-2020) timetable with the relative amount of visitors. It can be seen that despite the COVID-19 situation, there are still time-slots in the weekend which can be defined as 'busy'.

Table showing recent (as of 7-9-2020) relative amount of visitors. Green=quiet, yellow=little bit busier, red=busy. .

Due to the high amount of visitors, and the fact that the Heuvel Galerie is a closed, indoor facility with clearly defined entrances it could be a potential safety hazard. On busy days these entrances can become choke points and the hallways can become crowded.

The Heuvel Galerie could benefit from a system of the type which is developed in this project. Due to the five entrances (four street-level entrances and the parking lot entrance), the flow of people entering and leaving can be precisely monitored. The COVID-19 symptom detection system will be placed at all entrances. This way, the entire flow of people entering the Heuvel Galerie will be checked for COVID-19 symptoms. If a person is identified with possible COVID-19 symptoms, they can for example be asked to not acces the Heuvel Galerie and do a real test at a GGD testing location or be denied acces completely. This will create a more safe and pleasant shopping experience for the shoppers and a more safe working environment for the owners and personnel of the shops and restaurants.

Who are the users

User

The users of the robot are the people going in to the Heuvel Galerie, the medical staff that check for COVID-19 symptoms, the shop employees and the law enforcement.

The people going in to the Heuvel Galerie will be checked to see if they are coughing, if they are not, then there is no problem. However if they are coughing they could then be picked out by the medical staff and get a closer checkup. When the people are checked and it is seen that they do not have Covid-19 they would be allowed to enter. If they however do test positive, the law enforcement could keep these people out and with this prevent them from spreading the virus inside. For the shop-employees this is also very important since they come into contact with a lot of people. So having people screened at the entrance will lower the chance of them becoming sick and then also spreading it further.

Society

The main stakeholders for the society would be the government, the uninfected and infected people going into the Heuvel Galerie and the shop employees.

The local government is a stakeholder since they could spend money to place the device and are responsible for the local regulations around COVID-19.

The infected people are a stakeholder since they can be informed about having COVID-19. Which could change their behavior and limit the spread of COVID-19.

The uninfected people are a stakeholder since when there is a detection device for COVID-19 less people inside will be walking around infected with COVID-19, which results in a lower spread. For this reason the shop employees are also a stakeholder here, their working environment becomes a lot safer.

Enterprise

Since the robot should be developed the stakeholder would be the companies developing the product. The shops are also effected by placing the device at the entrance of the Heuvel Galerie, since this might prevent potential customers from being able to enter. On the other hand, this policy could also create potential customers for the shop since people might feel more safe shopping there compared to other places. The owner of the Heuvel Galerie is also a stakeholder since for them it is important to keep the shops happy so he will keep his revenue.

What are the users needs

Here the needs of the above identified users regarding the COVID-19 symptom detection system will be given. These requirements are partly based on a questionnaire conducted among shoppers at the Heuvel Galerie

User

People entering Heuvel Galerie

The requirements for uninfected shoppers are:

• Hindrance: Shoppers come to the Heuvel Galerie to have a fun day. The system should not cause excessive hindrance.

• Proper detection: Proper detection ensures possible COVID-19 carriers are identified so that others may shop in safety.

• Privacy: The system must comply with all privacy regulations. People prefer a crowd scanning system over a system in which they have to stand in a queue.

The requirements for infected shoppers are:

• Proper detection: The system should minimize the number of false positives and false negatives. People with good intentions that are unaware they carry the virus should be notified.

• Privacy: The system must be discrete since it can be embarrassing for people to be picked out for a COVID-19 test and it can cause unnecessary panic at the location. The privacy of the people being tested should be guaranteed such that for example bystanders cannot obtain the identity of the person being tested. Therefore, people have indicated to prefer a crowd scanning system over a queueing system.

Shop owners and employees

The requirements for the shop owners and employees are:

• Proper detection: Shop personnel will come into contact with a lot of different people while doing their work. Proper detection will benefit the safety of the personnel, for which the shop owner is responsible.

Medical staff

The requirements for the medical staff are:

• Ease of use: The system must be easy to use for the medical personnel. No special extra training should be required, a user manual should provide enough information to operate the system.

• Safe to use: While operating the system and working with people that possibly carry the virus, the medical personnel should not be exposed to too much risk of catching the virus themselves.

Law enforcement

The requirements for the law enforcement are:

• Proper detection: A proper detection ensures a safer work environment for the security guards of the Heuvel Galerie.

• Hindrance: The system should not be cause of any dangerous- or other unwanted situations requiring the security guard’s attention and thus creating more work for them.

Society

Local government

The requirements for the local government are:

• Increasing attractiveness: The system should create a safe shopping environment, making it more attractive for people from outside Eindhoven to go to Eindhoven and shop at the Heuvel Galerie, and possibly other stores/restaurants in the city.

People entering Heuvel Galerie

The requirements for the people entering the Heuvel Galerie as seen from a society standpoint are the same as seen from a user standpoint, except that when seen from the society standpoint, the requirement of proper detection becomes extra important because this will create safer shopping environments in general and could help reduce the number of infected people.

Shop owners and employees

The requirements for shop owners and employees as seen from a society standpoint are the same as seen from a user standpoint, except now again the proper detection requirement is extra important because this can lead to a healthier population in general meaning more potential customers.

Enterprise

Shop owners

The requirement for shop owners from an enterprise standpoint are:

• Hindrance: The system should not cause excessive hindrance, resulting in less potential customers visiting the Heuvel Galerie.

• Increasing attractiveness: The system should make it more attractive to visit the location compared to the situation in which the system was not in place or compared to other shopping locations in Eindhoven were no such system is installed.

Company developing the product

The requirements for the company developing the product are:

• Attractive product: The product has to be attractively priced in order to sell more units and make a profit. Therefore, it should not be too expensive to manufacture. It also needs to have decent looks to make it marketable.

• Comply with regulations: The manufacturer does not want to get into trouble for selling this product. It must comply with all regulations.

Owners of Heuvel Galerie

The requirements for the owners of the Heuvel Galerie are:

• Cost: The system should not be too costly to purchase and to operate.

• Comply with regulations: The owners of the Heuvel Galerie will the party operating the system via their employees. Therefore it is important that the system complies with all regulations such that the owners do not get into legal trouble.

As can be seen, for multiple stakeholders it is important that the product complies with regulations (health, privacy etc.) and that the system properly detects people with COVID-19 symptoms. This is one of the most important requirements. The above-mentioned user requirements will be converted to more concrete requirements for the design of the system in the RPC’s.

State of the art

Here, the State of the Art is given. It is divided into several topics. Each topic starts with a small summary of the State of the Art and after that the summarized articles concerning that topic are given.

Screening System for detecting COVID-19 symptoms and safety rules compliance

To verify the feasibility of an automatic system which screens subjects for COVID-19 virus using cough sounds some research needed to be done. Several research topics have been investigated such as: COVID-19 detection trough cough sound, conventional techniques to diagnose COVID-19, thermal imagery for mass fever screening and infrared technology for temperature sensing in humans.

An alternative method to screen subjects for COVID-19 would be through cough samples. Currently, research is being done by a couple of groups in the UK and USA. Resulting high accuracy of the algorithm has encouraged a mass collection of cough samples, even through the use of mobile APPs.

General guidelines and techniques to diagnose the COVID-19 virus. Some of the discussed techniques for diagnosis are molecular tests (PCR and serologic test), CT scans and emerging protein tests.

Several articles discussing the efficacy of infrared skin thermometers and infrared imaging in COVID-19 screening are presented. The conclusion is that handheld infrared thermometers have a low sensitivity and should not be used as a screening system. While discussions arise when assessing the usefulness of mass fever screening using thermal imaging. Many factors need to be taken into account such as: camera resolution, ambient temperature, distance from the camera and even marketing.

Articles

AI4COVID-19: AI enabled preliminary diagnosis for COVID-19 from cough samples via an app. (A. Imran, et al. 2020). Summarized by Ikira.

This paper discusses an alternate way of testing for covid-19, since the current testing:

- Requires people to move through public places which could spread the virus

- Puts medical staff at risk for the personal

- Costs a lot

- Not everyone goes to a medical facility to get tested

The solution they came up with is to make an smartphone app that does a preliminary test based on your cough to see if you have covid-19 or not.

The reason for choosing this method is because:

- Prior studies have shown different respiratory syndromes have distinct features

- Coughing is the way the virus is mostly spread.

- There are less medical conditions that cause cough than fever

The steps taken for the making of the smartphone app are to first test if covid-19 has specific coughing characteristics (Results show there are). Second to train an engine that can distinguish a cough from background noise. third to train an engine that can distinguish between the different types of coughs and fourth to check the feasibility of the app. The accuracy of the proposed app is encouraging enough to support large scale testing.

Exploring Automatic Diagnosis of COVID-19from Crowdsourced Respiratory Sound Data (C. Brown, et al. 2020).

Very similar research project as , but conducted by the University of Cambridge. Their research also tries to detect COVID-19 trough cough noises and respiration. Their research differs on two aspects. First, this study only uses crowdsourced audio recordings, while uses data recorded in controlled environments. Secondly, this study uses simple machine learning (SVM) while use a end-to-end deep learning model, which often overfit results when small samples are used.

Semantics and types of cough. (K. F. Chung, et al., 2009)

Semantics of cough was very confusing and unorganized since also classification of cough type was dependent from methods used and form clinic-to-clinic.

Definition of cough

• First definition: Three-phase expulsive motor act:

1. Inspiratory effort (inspiratory phase)
2. Forced expiratory effort (compressive phase)
3. Rapid airflow (expulsive phase)

• Forced expiratory manoeuvre against a closed glottis producing a sound

Types of cough

Coughs identified by their cause (e.g. asthma, reflux). Coughs are defined by their characteristics (e.g. wet, moist, chesty). Coughs can be acute (i.e. <3 weeks), subacute (i.e. 3-8 weeks) or chronic (i.e. >8 weeks). Coughs can occur in isolated episodes or in epochs. See Table 1.

Clinical analysis of coughs

Only quantitative data available is cough count. Coughs can be counted using automated audiovisual recordings. Automated cough counting does not distinguish between isolated and epoch coughs. More sophisticated analysis which measure pressure, airflows and abdominal muscle electromyograms can give a quantitative intensity of the cough. But this analysis is better suited for a research clinic instead of an investigative clinic.

Estimated effectiveness of symptom and risk screening to prevent the spread of COVID-19 By (K. M. Gostic, et al. 2020). Summarized by Lieke. This article uses a mathematical model about the spread of emerging pathogens from 2015 combined with the current knowledge about the epidemiological parameters of the COVID-19 spread to estimate the extent in which traveler screening will be useful. Because COVID-19 has a long incubation period many cases are fundamentally undetectable, which will lead to screening missing more than half of the infected people. This number might increase a bit when the epidemic is stable instead of growing, because the times since exposure will be more evenly distributed.

Real-time tracking of self-reported symptoms to predict potential COVID-19 (C. Menni, et al. 2020). Summarized by Lieke.

Symptoms and indicators of COVID-19 include loss of taste and smell, anosmia, fatigue, persistent cough and loss of appetite.

Asymptomatic Transmission, the Achilles’ Heel of Current Strategies to Control Covid-19 (M. Gandhi, et al. 2020). Summarized by Lieke. Infection control relies heavily on early detection, but because COVID-19 has a long incubation time this does not work. This is why it is important to continuously test especially nurses on COVID-19, because it can be detected before the symptoms start to manifest.

Crowds by Example (A. Lerner, Y. Chrysanthou, D. Lischinski, 2007). Summarized by Lieke.

Crowds can be modeled in several different ways, and in this study the people are tracked using video input, and modeled from that. Afterwards, the pedestrians were asked what influenced their path and to what extent. Using this information a model can be made. The influence factors are, to name a few, the topography of the terrain, the presence of obstacles, proximity to other people, the pedestrian’s personality and state of mind. Setting up some equations based on the experiment a model can be made that is able to simulate a crowd.

Diagnosing COVID-19: The disease and tools for detection (B. Udugama, et al. 2020). Summarized by Lieke. To detect COVID-19 in the hospital there are several tests that can be used. Molecular tests are useful because they can target and identify specific pathogens, although nucleic acid testing is the primary way of testing for COVID-19. CT scans can be used but have a lower specificity because the imaging features overlap with other viral pneumonia. Emerging diagnostic tests for COVID-19 are protein tests, more developed nucleic acid tests and detection at the point-of-care, such as lateral flow antigen detection and microfluidic devices.

Detecting the Coronavirus (COVID-19) (P. Pokhrel, et al. 2020). Summarized by Ikira. In December 2019 a new virus was discovered, the SARS-CoV-2. This virus causes the disease COVID-19 which has a medium reproduction rate(2.25%) and a medium mortality rate (5.7%). However contrary to the looks of only being a mild virus, when it remained untreated the sick people began to overwhelm the healthcare system which spread panic. SARS has a lot of similarities with COVID-19 and was eradicated in 2003 by syndromic surveillance, isolation of patients and not physical contact. If this will also work for COVID-19 will remain to be seen.

The characteristics of COVID-19 appear 3-7 days after infection and include fever, coughing and body weakness. People of old age or with underlaying illnesses have a higher mortality rate. People of ages are susceptible to SARS-CoV-2 which transmits via contact or bodily fluids.

One of the most important ways to limit the spread is by early detection so precautions can be taken. The current ways of detection COVID-19 are via a CT-scan, Nucleic Acid based methods and Immunoassays.

CT-scans contain specific characteristics of COVID-19 which helps to identify if the person has COVID-19, however these characteristics may overlap with other infections and give false positives. The test is more accurate if the test is executed during specific stages.

Nucleic Acid based method is a way of detection of RNA viruses in a clinical setting. Based on this method the CDC developed a real time RT-PCR diagnostic unit. Although this method is less sensitive than the CT imaging at certain stages, it specificity is much better making it a better detection method.

Immunoassay detects viral protein antigens or serum antibodies of people exposed to SARS-CoV-2, this method can detect the antibodies 3 days after infection it however also ha a lot of drawbacks.

The now emerging detection techniques are isothermal amplification for nucleic acid targets and lateral flow based detection of nucleic acids an proteins.

Comparison of Non-Contact Infrared Skin Thermometers (T. Fletcher, et al. 2018). Summarized by Ikira.

Non-contact infrared skin thermometers are being used more as a means to asses body core temperature and signs of diabetic foot ulcerations In the paper 3 different thermometers were tested. All the manufacturers claimed less than 0.3 °C in the 22-40 °C range.

For infrared measuring devices there are changes in the measured temperature depending on the size of the target. This is also called the size-of-the source effect. Something else that influences the measurements is the difference in distance or the target from the measuring device. This is important since for these measuring devices there is no way to have the exact same distance each time and is thus included in testing these devices.

The results of this paper show that the tested measurement equipment did not come close to the before specified 0.3 °C and could reach up to 6 °C. The paper furthermore showed that the size-of-source effect and the distance matter a lot in the accuracy of the measurement device.

Comparison of Infrared Thermal Detection Systems for mass fever screening in a tropical healthcare setting (M. R. Tay, et al. 2015). Summarized by Ikira.

Fever screening detection systems such as infrared thermal detection systems (ITDS) have been used before for rapid identification of potential cases of infected people. ITDS use the difference in the ambient temperature and the temperature of the subject.

Three different ITDS were used in Singapore to check if people had a fever. For these people an oral check was taken as a reference. In the paper sensitivity is described as people that have a fever are also detected as having a fever and specificity is described as people without a fever are also not seen as having a fever. Of the three tested ITDS they all had high specificity , 2 showed low to medium sensitivity while one showed high sensitivity. The ones that showed the lower sensitivity were older models while the newest model showed the higher sensitivity.

Crowd temp screening is a joke, says one expert (K. Field, 2020)

Accuracy of the thermal camera depends on several factors (including human variability) and in ideal conditions the maximum accuracy is of ±2 deg C. (which to detect feverish patients is not enough).

The human body is a living and closed loop system it will fluctuate its temperature on various conditions (i.e. ambient temperature, energy expended, and alcohol consumption).

According to FDA guidelines the ideal part of the face to detect accurate temperature would be the inner canthus (tear duct). To measure the temperature of such a small area needs at least 3 x 3 pixels (5 x 5 for better results). Therefore to film such a small detail in a crowd environment a thermal camera should have a resolution of at least a million of pixels (around 12 mega pixels) which do not yet exist.

"Are you just deploying this gear for ‘feel good’ needs, or are you truly looking to measure and eliminate feverish people?"

"I could deploy a doctor that says he can tell your temperature is not normal by looking at your nose--and that might be just as accurate"

The expert's opinion: If you can’t do it right--don’t bother.

Remote monitoring of breathing dynamics using infrared thermography (C. B. Pereira, et al., 2015)

Technique using thermal imaging in combination with a robust algorithm to detect breathing and to measure breathing rate (BR). The algorithm is a able to compensate for the movement of the subject.

The system shows accurate estimations of the breathing rate even under non ideal conditions (i.e. movement of the subject). For accurate measurements, the ambient temperature must be controlled. For further improvement of breath detection movement of the shoulders should be included in the algorithm.

Privacy Issues with Smart Technologies

Some research on privacy issues was done as a possible alternative idea. Nevertheless, privacy issues do tie quite strongly to the current subject. Since people are being recorded both on video and audio the question arises, what is done with the data? A more in depth research on ethical issues could be conducted. Some of the article discuss ethical use of surveillance technology, cyber security smart camera networks and in smart homes.

Articles

A Systematic Approach Towards User-Centric Privacy and Security for Smart Camera Networks (T. Winkler, et al. 2010). Summarized by Ikira.

Fully digitized systems analyze video footage and sets of an alarm if there is suspicious activity. There a however still issues of privacy for this. Since the footage is stored, people can retrieve it and study the behavior of people. The current solutions for this are to blur out faces or to extract encrypt sensitive image regions.

Desirable properties of a secure an privacy-preserving camera system would be:

- Detection of sensitive regions

- Blanking

- Obfuscation and scrambling

- Abstraction

- Encryption

- Multiple privacy levels

For user acceptance it is important that the user gives consent to being filmed and can control it when they are being filmed by for instance placing signboard that the area is being filmed. Furthermore it is also important that the users can give feedback.

The paper sketches concept of a level approach for trustworthy smart camera’s. As a basis for the camera there will be hardware security around a microchip called the trusted platform module. To support the different privacy levels, motion detection is used. The lowest level of security is where only a background is visible while the person and their immediate surroundings are cut out by a black box. The next level would is done with edge detection, here only the person himself is visible and nothing of the surroundings. This is followed by the last level where the person and their immediate surroundings is visible. In other words only the black box that was cut out for the lowest level is visible.

A Smart Home is No Castle: Privacy Vulnerabilities of Encrypted IoT Traffic (N. Apthorpe, et al. 2017). Summarized by Ikira.

Smart Home devices collect data not only when you are on the internet, but also about your daily life. There is a privacy concern if a passive network observer deduce behavior in the change in internet traffic. Some people say that due to encryption no sensitive information can be seen, however it can be argued that meta data and traffic patterns can reveal sensitive information. In this paper a strategy to recognize user behavior with traffic load of internet of things devices is given. Four Smart home devices are used for this. From all of these devices a network observer could infer behavioral patterns from observing the traffic load. Thus it would be necessary for devices to mask the true rate of devices. For the research in this paper a few assumptions were made. First of it is assumed there is only a passive network threat. Second packet contents are not used and finally it is assumed that the observer can obtain and study those devices. The way to analyze these traffic rates is to first separate traffic into sperate packet streams, then to find which stream belongs to which device. With this the traffic rates can be examined and the behavior of the consumer can be deduced.

All Eyes on You! Impact of Location, Camera Type, and Privacy-Security-Trade-off on the Acceptance of Surveillance Technologies. Summarize by Ikira.

Crime surveillance technologies(CST) have increased in the past years. Critics fear that the recording and storing if the data will invade the privacy of the people. The acceptance of the people for these CST are important. This paper discusses the acceptance of the people towards different types and locations of the CST, at which turning points acceptation becomes rejection, the trade of between safety and privacy. To research this questionnaires were taken.

The results show that people find the locations of the CST the most important followed by the crime reduction, handling of the recorded data and last the camera types. The paper further shows based on the before named subjects the how much people accept it the CST. The most rejected was surveillance in your own home. To research the trade of between security and privacy 3 scenarios were made, 1 with high security low privacy, 1 with average security and average privacy and 1 with low security and high privacy. The high security came out the best of these scenarios.

Autonomous robots against COVID-19

A prime example of a (USE) robotic technology in use to combat the COVID-19 pandemic. This article describes the successful employment of a robot using UV lamps to disinfect hospital rooms.

Autonomous Robots Are Helping Kill Coronavirus in Hospitals (E. Ackerman, 2020). Summarized by Lieke.

UVD Robots is making robots that are mobile arrays of powerful UVC lights that kill microorganisms. UV lights on carts are used to disinfect areas for some time, but they are dangerous for human skin and when they are operated by humans it is prone to human error and it might skip a few places. This newly developed product is able to autonomously disinfect a room between 10 and 15 minutes, with the robot spending 1 or 2 minutes in five or six different positions to maximize the surface it disinfects.

General robotic technologies in the medical field

Since at the beginning of the project the topic was unclear some research was done on general technologies used now and applicable in the near future.

Articles

Can robots handle your healthcare? (N. Davies, 2016). Summarized by Wout.

The demand for healthcare robots will rise due to an ageing population and a shortage of carers. Japan is a leading country in the field of healthcare robots. They develop their robots in close coordination with elderly and patients.

Safety is an important aspect for healthcare robots. They must be absolutely fool proof. In order to safely operate around humans, robots need to feature things like quick reflexes, communication capabilities and interaction capabilities via video and audio. This translates into a wide range of sensors, cameras and actuators the robot needs to have. Next to a robot’s features, their looks are also important. Robots that try to impersonate actual humans, but still show imperfections, can make users uneasy.

An area in which robots can do a better job than humans is being consistent in the quality of the care that is provided. After a long difficult shift, humans can be prone to errors while robots are not affected by exhaustion or other emotions.

For simple routine healthcare tasks, robots can be a good solution.

How medical robots will change healthcare (P. B. Nichol, 2016). Summarized by Wout.

A possible major field in future healthcare is medical nanotechnology, and more specific nanorobots. These robots will work inside the patient at cellular level. The most important stakeholders in robotic healthcare are not the people that develop the technology, but the people that will use it.

There are three main ways in which robotics can be deployed in healthcare: Direct patient care (e.g. robots used in surgeries), Indirect patient care (e.g. robots delivering medicine), Home healthcare (e.g. robots that keep the elderly company).

Acceptance of Healthcare Robots for the Older Population: Review and Future Directions (E. Broadbent, et al. 2009). Summarized by Wout.

It is projected that the proportion of people older than the age of 60 will double between 2000 and 2050. This increases the need for healthcare. Alongside this there is also a shortage of people working in the healthcare industry. In western countries, most of the older people want to remain living independently in their own homes for as long as possible. To facilitate this, smart solutions need to be found, for example healthcare robots.

Of great importance is the acceptance of healthcare robots by the people in need of care. There are three requirements for acceptance of a robot to occur: motivation for using the robot, sufficient user-friendliness and a feeling of comfort with the robot.

Healthcare robots need to be able to adapt to individual differences since not all patients are equal.

Basic healthcare robots have been developed which can assist with simple tasks like: assist in walking, reminding of doctors appointments, carry objects, lift patients, providing companionship. Mitsubishi sold the first commercially available mobile robot in 2005. This robot is designed to help in the home.

The older the age, the less willing people are to use robots. Older people are more fearful of new technology. However, they are more willing to accept assistive devices if this helps them maintain their independence.

Studies have shown that people prefer robots that do not have a human-like appearance. For companionship robots an animal-like appearance is preferred. Other examples of desired characteristics of a robot for the elderly are slow moving, safe, reliable, small and easy to use. Older people also appreciate having some form of control over the robot because this reinforces their sense of independence.

Medical Robots: Current Systems and Research Directions (R. A. Beasley, 2012). Summarized by Wout.

An example of a great success in medical robotics is the surgical robot da Vinci. The biggest impact of medical robots has been improving surgeries that require great precision.

Medical robots can be employed during brain surgeries. By using medical images (e.g. CT-scan) a robotic arm can orient the tools in the correct direction, or the surgeon can specify a target which the robot guides the instrument towards with submillimetre accuracy. Medical robots are also used to assist in spine surgery.

Another application field of medical robots is orthopaedics (e.g. hip or knee replacement). An example of such a robot is Robodoc, which is used for milling automatically according to a surgical plan. Other devices, like the RIO, are designed such that the robot and the surgeon hold the tool simultaneously. The RIO ensures that the surgeon makes the correct movements.

There also exist devices which are controlled by the surgeon (technically not robots). The surgeon uses joysticks to manipulate the tools with great precision. An example of such machines are the Zeus and the da Vinci. These systems filter out hand tremors and enable the surgeon and patient to not be at the same location.

InnoMotion is a robot arm which can accurately guide a needle using CT or MRI imaging.

Robots exist that can help a surgeon guide a tube through a patient’s blood vessel. The surgeon can steer the tube and force feedback sensors tell the surgeon how much pressure is being applied.

Robotic arms are also used in radiosurgery, in which a specific area inside the patient is targeted with beams of ionizing radiation from multiple angles.

Robots are also used to perform CPR. A band is placed around the patient’s chest which is pulled tight by an actuator to compress the chest. The tightness of the band is adjusted based on the patient’s chest size.

Microprocessor-controlled prosthetics are prosthetics in which the patient is assisted in the movement or are controlled by the patient to perform movements which have been lost (e.g. exoskeletons).

Another area of medical robots consists of assistive and rehabilitation systems. These include things like a controllable arm holding a spoon or a grasper which can be attached to a wheelchair.

In the future, medical robots will become smaller and cheaper. More functionality will be added. Also, robots will be used more for medical training. Medical robots need to provide solutions to real problems, otherwise there is a change that they are displaced by other medical advancements.

Medical robots go soft (E. Strickland, 2017). Summarized by Wout.

New kinds of medical robots are being developed of which the parts that touch the patient are soft. This is to prevent damage to organs and blood vessels and makes sure the human body does not surround the object with scar tissue which can diminish its effectiveness.

An example of a soft medical robot is a sleeve that surrounds the hart. This sleeve contains rings which can be pneumatically inflated to detect and force a heartbeat in a certain tempo.

Another example is a device which can be implemented in the body to give of small dosages of a drug. This device is made of a soft gel with a mechanism inside which can be activated from outside the body.

A final example is a soft gripper-hand made of hydrogel. The hand can be actuated by forcing water into the hand. This hand can be placed on a robotic arm and used for grabbing organs inside the human body without damaging them.

Robots in Health and Social Care: A Complementary Technology to Home Care and Telehealthcare? (T. Dahl, et al. 2013). Summarized by Lieke.

This article discusses several specific robotics applications.The robots that were considered were classified in 9 categories: Surgery robots, Robots supporting Human-Robot interaction tasks, Robots providing logistics in care home environments, Telepresence and companion robots, Humanoid robots for entertaining and educating children with special needs, Robots as motivational coaches, Home assistance robots for an ageing society, Human-Robot relationships in medical care and society, and Human- Robot relationships in medical and care situations.

Between these categories, there are a few ingredients for success that can be found. The general ingredients for success include: Adequate level of personalization, Appropriate safety levels, Proper object manipulation and navigation in unstructured environments, Patient/user safety, Reliability and robustness and Sustainability. In addition, robots that serve motivational or social purposes also need to be personable, intelligent and highly interactive.

However, there are also some concerns with using robots extensively. For example, humanoid robots need to have a set-up etiquette, like following the eyes of a talking person. What also needs to be kept in mind, especially using social robots with elderly people, is that the extensive use of robots could lead to a potential reduction of human contact, social isolation, an increasing feeling of objectification and a loss of privacy. Furthermore, cost effectiveness plays a great deal in the practical use of robots.

Multimedia Companion for Patients in Minimally Conscious State (MCS)

Novel imaging techniques have enabled communication with minimally conscious patients. The idea is to create a multimedia system for patients in MCS. This system would project films, music and videos of family members. This system can communicate with the patient only with binary (yes or no) questions. Communication is used to let the patient navigate the multimedia system (i.e. choosing what to watch or do). The yes or no answers are given by the patient's ability to modulate brain activity. The patient's response is decoded by sensing which areas of the brain intensify in activity.

System:"Do you want to watch the news? If YES imagine playing tennis. If NO imagine walking in the rooms of your house."

Patient: The patient imagines to play tennis.

The motor area of the patient's brain activates.

Activity is recorded by the system.

The system plays the latest bulletin.

Articles

- First technique using functional Magnetic Resonance Imaging (fMRI) and PET.

The Aware Mind in the Motionless Body (M. M. Monti, A. M. Owen, 2010) & Detecting Awareness in the Vegetative State (A. Owen et al., 2006)

Case Study: Martha

Martha, a young woman of not even 30 years old is victim of a car accident. She survives the accident and slips into coma; her eyes are closed, her body is motionless and unable to breath autonomously. After a week her eyes open, her body is able to breathe again autonomously and even makes small movements. For example, she yawns, she stretches her arms and slightly moves her body in her bed. It seems she is even able to sleep and wake up.

Vegetative State, Comatose state, and sleep

When a patient in a comatose state is able to sleep and wake up again, she/he is said to be in the Vegetative State (VS). A patient in a vegetative state might have his/her eyes open, can breathe autonomously, can move his/her body, can produce urine, and can vocalize sounds. The big question, are patients in vegetative state conscious? Consciousness is made of wakefulness and awareness. A healthy person is both awake and aware, when not asleep. A comatose patient is neither awake nor aware. While a patient in vegetative state is awake, but is not aware. Now around 40% of patients diagnosed in the vegetative state seem to be aware. Therefore, these misdiagnosed patients are said to be in a Minimally Conscious State (MCS). Often while diagnosing patients the “Lack of evidence of consciousness is, in this situation, equated to evidence of lack of consciousness” (Monti, Owen, 2008)(i.e. False Negative). To distinguish patients in MCS from VS one of three factors must be met:

• Awareness of the environment

• Sustained, reproducible, purposeful or voluntary response to visual or tactile stimuli

• Language comprehension and/or expression

Novel neuroimaging techniques are used to detect markers of consciousness, such as Positron Emission Tomography (PET) and functional Magnetic Resonance Imaging (fMRI).

Experiment 1: brain response to speech of patients in VS

Two patients in VS showed brain activities similar to healthy volunteers in response to speech and picture of faces. The two patients in VS listened to simple and complex (semantically ambiguous) sentences. The complex sentences activated, in the VS patients, the same brain area as in healthy volunteers. Furthermore, similar experiments were conducted with healthy volunteers heavily sedated. In the heavily sedated volunteers, no activation of the same area was recorded. This shows a higher level of awareness (some language comprehension) in the VS patients than the sedated volunteers. Nevertheless, simple brain responses cannot be taken as a factors of awareness nor consciousness. Much of sensory processing is automatic. Such as lines depicting a face or speech of a familiar language. Therefore, the type of stimuli and the type of brain activity recorded must be attentively chosen.

Experiment 2: voluntary modulation of brain activity in VS patient

A young woman victim of a car accident slips into coma, then wakes up again, but is diagnosed in the Vegetative State. The patient was given spoken instruction. The first instruction was to imagine playing tennis. The second, to imagine walking in her own apartment. When the patient was instructed to imagine playing tennis intense activity in the motor area of the brain was recorded. While when she was asked to imagine walking in her apartment activity in the parahippocampal gyrus (memory retrieval area), posterior parietal cortex (spatial reasoning area) and some motor areas. Healthy control volunteers, given the same instructions, reproduce the same brain activity of the VS patient. Unless the volunteers did not understand the instructions or decided to not comply with them. Therefore, in this experiment the idea of an automatic brain response is discredited.

- Second technique using Transcranial Magnetic Stimulation in combination with high density Electroencephalography (TMS/hd-EEG). This technique is much more practical and portable than fMRI or PET.

A perturbational approach for evaluating the brain’s capacity for consciousness (M. Massimini et al. 2009)

It does not confirm if a person is conscious or not. The technique weighs the capacity of the brain for consciousness. The technique does not require any input from the user. In this article the technique has only been used to differentiate brain response, to TMS excitation, in three conscious states: wakefulness, long-wave sleep, and REM sleep. More research must be done with sedated patients , and patients with brain injuries (i.e. Minimally Conscious State or Vegetative State).

1. Start from a theory describing how consciousness is created. Integrated Information Theory of Consciousness (IITC) suggests that consciousness is created by a physical system.

2. Measure the brain’s capacity to generate consciousness. Measurement is done by poking (exciting) the concerned physical systems, with TMS, and record the response, with hd-EEG.

Results: Brain responses to TMS during wakefulness are similar to those during REM sleep, showing effective connectivity between brain areas (i.e. the stimuli travel between various regions). While long wave sleep shows loss of cortical integration (i.e. the stimuli remains local).

Similar systems

Here, an overview of similar COVID-19 (or other disease) symptom detection systems are given.

SARS-2/COVID-19 checkpoint from Unival group: https://www.mining-technology.com/sponsored/screening-and-checkpoints-in-the-age-of-covid-19-how-to-safely-re-open-your-project/

Airport Security Screenings:https://www.h4-solutions.com/blog/thoughts-on-airport-security-design-checkpoint-construction-practices

USA TSA design: https://acconline.org/documents/20140507-000_Checkpoint%20Design%20Guide%20(CDG)%20Rev%205.1.pdf

Article: https://www.h4-solutions.com/blog/thoughts-on-airport-security-design-checkpoint-construction-practices

EU accelerated project: https://cordis.europa.eu/project/id/285311

Cough detection Camera

Developed by the Center for Noise Control at Korea Advanced Institute of Science and Technology (KAIST). The system is a combination of an array of microphones and a video camera. The system uses a deep learning algorithm to distinguish cough from other sounds. The algorithm can recognize the sound of coughing even with some level of background noise. The array of microphones detects the position of the cough and maps it to the video feed of the camera. To detect the position of the sound source Time Difference Of Arrival is probably used, but not specified. After detection the subject coughing is tracked, and the number of coughs is counted. This system could be useful for analytical purposes. As a tool for a statistical study to measure the number of people coughing and how many times they cough. Also, it could be used as a monitoring device in hospitals to help for diagnosis. No scientific article or report of the project could be found online. Therefore, no quantitative data nor specifications of the device could be found (e.g. max sensing distance, field of view, resolution, Signal to Noise Ratio, etc.).

Social Distancing Software

Cloud based AI which takes existing camera feed and alerts on social distancing, mask detection, gun recognition/active shooter. The system has various features such as:

• Social distancing: calculates distances between humans and mask detection.

• Threat detection: weapon identification, face recognition.

• Smart Building: Building management: People flow, loitering detection.

The system is said to be budget friendly because only a software license is bought. The system takes all the input from already existing infrastructure. The system does not require to build extra expensive infrastructure for processing or for storing data since it is cloud based. Unfortunately, this design feature raises some issues. All the sensitive data is processed, stored, and shared by and with a third party which may arise security concerns. For example: Where is the data processed? Is the data stored on the cloud even if the client denies? The system boasts it is privacy conscious and bias free. Further research should be done if interested.

Temperature Screening

IMPORTANT:

• Ambient temperature (20-24 °C) / humidity (10-50%)
• Non-reflective IR background
• No moving air
• No direct sunlight
• No radiant heat sources (generators, etc.)
• Black body. Placed on same plane as subject’s face. Camera perpendicular to black body.
• Proper lens with proper distance

Type of screening detection systems:

• Handheld Infrared thermometer
  • Extech IR200 (from FLIR)
  • Shenzhen WTYD Technology Limited HT-F03B (China) tested by Singapore study
• Infrared imaging systems
  • HIKVISION (Brussels)
  • Omnisense (Singapore)
  • FLIR / MoviTHERM
    • FLIR Screen-EST mode. Integrated software that measures for high body temperature. Does not require black body (i.e. it is not an absolute temperature measurement).
  • Thermal Screening Solutions
  • Satir

Picture
Model	HIKVISION DS-2TP21B	FLIR E54/86 EST	FLIR T540/60 EST	FLIR A500/700 EST	FLIR A320	HIKVISION DS-2TD2617B/36B	Satir CK350-F	Omnisense Sentry MK4	Seek Thermal Seek Scan
IR Resolution (HxV)	160×120px	384×288 / 640×480px	320×240 / 464×348px	464×348 / 640×480px	320×240px	160×120 / 384×288px	384×288px	384×288px	206×156px
Thermal accuracy	±0.5°C	±0.3°C @ 30-45°C			±0.5°C @ 37°C blacbody	±0.5°C, (±0.3°C blackbody)	≤0.3 ℃	±0.1°C @ 30-45°C blackbody	±0.3°C @ 30-45°C blackbody
Thermal Sensitivity NETD	<40mk @ 25°C	<40mK @ 30°C (24° lens), <30mK @ 30°C(42° lens)			50mK @ 30°C	≤40mK @ 25°C	40mK @ 26.85°C	≤50mK nominal (probably @30°C)	40mK @ 30°C <50mK @ 25°C
Field of View (H×V)	25°×18.7° (6.2mm lens)	14°×10° (29mm lens), 24°×18° (17mm lens), 42°×32° (10mm lens)			25°×18.8° (18 mm lens)	25°×18.7° (6.2mm) / 28.8°×21.6° (13mm)	46°×35.3° (8mm)	24.6°(Horizontal)	35°×26° (4mm)
Fixed Lens	Yes	Interchangeable			Yes	Yes (both)	Yes	Yes	Yes
Automated Screening Software	Free Central management software	Included			Free Central management software	Free Central management software (iVMS-4200)	NA	Included	Included (NOT Windows PC)
Features	Integrated display, tracking people	Integrated display,	Integrated display,	Alarm Functions, motorized focus	Used 2003 SARS,	iVMS-4200 software features AI detection,	Motion detection, Body recognition, Temperature Alarm	Used 2003 SARS, Integrated display, Alarm Audiovisual with intelligent febrile tracking, with blackbody	Alarm audiovisual, with blackbody
Scan time per person	NA	"Seconds"(with Proprietary software)			NA	NA	NA	"Seconds"	1-3s

Crowd Monitoring

The system calculates the number of people in indoor and outdoor spaces. This system uses a frequency scanner placed at waste level. Scans crowd capacity by recording the Received Signal Strength. The signal attenuates depending on the quantity of people present. The system is completely modular. Each unit is made of a transmitter and receiver. Many units are combined in a network. The received signal is measured from each unit and a model is created on the number of people. Flow and flux of people is analyzed in a totally anonymous way. The main features of this system are its accuracy, privacy friendly, real-time measurement capabilities and no radio frequency interference (RFI).

Privacy issues and regulations

Here, a study will be done on the possible privacy issues and legal issues arising from a system as proposed in this project.

Privacy issues

What is privacy?

Before the issues of privacy can be discussed, it is important to know what exactly privacy is.

There are a lot of different definitions for privacy, according to Warren and Brandeis it is “The right to be left alone” [35]. Others define it as a way of information control.[36]. Another description of privacy is “The condition of not having undocumented personal knowledge possessed by others”[37] William Prosser Separates privacy in 4 different categories:

1. Intrusion upon the plaintiff’s seclusion or solitude, or into his private affairs.

2. Public disclosure of embarrassing private facts about the plaintiff

3. Publicity which places the plaintiff in a false light in the public eye

4. Appropriation, for the defendant’s advantage, of the plaintiff’s name or likeness.[38]

For the case of privacy for the device we are making the first point here is what we take into account when we are talking about privacy.

Constitution

The constitution also states rules about privacy and they way they are processed. These rules are in article 10, 11, 12 and 13 of the constitution[39].

Article 10 states[40]:

• Everyone has the right to privacy for their private life, the government is not allowed to use personal information for no reason and everyone is allowed to view their personal information and is allowed to change them if they are incorrect.

Article 11 states[41]:

• Everybody has the right to decide for themselves what happens to their body. Nobody is allowed to forcefully do medical test to their body. There are exceptions if this is regulated by the law.

Article 12 states[42]:

• Without permission of the resident entering the residence is only allowed with a statutory exception. This law also apply for workspace or public space.

Article 13 states[43]:

• The government is not allowed to read or listen to correspondence between citizens except in cases where the law allows it and by order of the judge.

For the device we are making and the procedure with using it these laws influence it by,

Article 10:

• None of the data that was collected will be saved and after it is used to detect the person the medical staff needs to pick out it will be deleted. Also no personal information is asked during the process. The device is not allowed to be used to collect any personal information.

Article 11:

• The extra tests can not be forced to be done to people if they are initially detected as having a symptom of COVID-19.

Article 12:

• It is allowed to refuse entry if someone is tested positive for COVID-19 or refuses to have further tests for COVID-19.

Article 13:

• This article does not have any effect on the device or procedure being done.

Case corona-app

One of the more known subjects in the news is the privacy issues regarding the corona-app. The app would track if you came into contact with someone who has corona and if you were close enough for a long enough time it would send you a message indicating this.[44] On 19-4-2020 a privacy analysis was done for 7 corona-apps. [45] In this analysis the principles that should be met were:

• The data that is used should not be able to be led back to anyone (anonymity)

• As little as possible false positives(correctness)

• Data can exclusively be shared if either there is contact- or source research or if the user gave his/her consent.

• The use should be aimed at simplifying contact research. (purpose limitation)

• There is an information portal where users can report mistakes and vulnerabilities. (transparency)

• Conventional security standards

• If the app is not effective or necessary any more data must be able to be deleted.(removal)

• Comply with the Algemene verordening gegevensbescherming(AVG)

This a analysis however showed that none of these apps meet the requirements stated by the AVG.

A group of privacy advocates wants to have a few firm commitments [46] namely :

• The app is only allowed to have 1 goal

• It must be based on scientific insight and proven effective

• Proven reliable and based on expertise.

• The application of the application is by definition temporary.

• Not traceable to individuals.

• Use as little data as possible.

• Do not store personal data centrally.

• Safe and resistant to abuse.

• User-friendly and accessible.

• Never under pressure from the government and third parties.

Users

A detection system that checks if people potentially have COVID-19 can have a lot of privacy issues, after all to detect this a lot of sensors are needed. The issue due to privacy violation is the fact that the habit of people can be deduced from the information seen or saved by the detection system. To prevent this steps should be taken in the design to prevent this issue.

According to T. Winkler, et al. a few desirable properties of a secure an privacy-preserving camera system would be: Detection of sensitive regions, Blanking, Obfuscation, Encryption, Multiple privacy levels, User consent, User feedback.

These are also points we try to implement in our design as well.

• Detection of sensitive regions: the detection system will only be installed at the entrance of the Heuvel Galerie and aimed at people entering.

• Blanking: When there are visual images involved, the background will be cut out so only the people/person involved will be visible.

• Obfuscation: First an infrared camera will be used. With this it is impossible to identify a person. Only when someone is detected to cough will the other camera be turned on.

• Multiple privacy levels: The system has multiple privacy levels by first only using sound and infrared measurements to detect if some is coughing. Only when it is necessary will the level go up to show who the person was that coughed. Here only the person himself is shown and not his surroundings to prevent any additional information from being seen.

• User consent: A board that shows what the detection system does will be made so that people know that the detection is there and what it is for. If the people do not consent to this type of detection system, they could also go somewhere else.

Questionnaire

From the questionnaire it can be seen that most people do not think this device intrudes their privacy. During the questioning of the people it became clear that most people do not really care about the information that could be collected by it and possibly sent to other parties.

Privacy general

https://www.sciencedirect.com/science/article/pii/S1094996819301124?casa_token=VJqlJNVakaAAAAAA:Rf6pm2d-XWnVqu-lByhp-H30Mw6x-suKx7FEuo462rF1nHJKsDKmI0aXT7Oj6QIG0-cC5RUTdBfR

https://www.aeaweb.org/articles?id=10.1257/jel.54.2.442

https://www.sciencedirect.com/science/article/pii/S0736585317303180?casa_token=BX4duXoM51oAAAAA:Puv2Sn8SWK4WYbF2E69-OPsEyo4lthvbPi3kxHgYhunX3eCRFPmxqBbt_zRHp9Cy5JmQeohWK_Qq

https://www.researchgate.net/publication/228322650_Privacy_Its_Meaning_and_Value

https://link.springer.com/content/pdf/10.1007/s10676-012-9291-0.pdf

Legality of doing health checks

A question that arises when a COVID-19 symptom check, followed by a possible temperature measurement and cough-test is performed in a public place, is the legality of such a procedure.

In the proposed design, a system will first detect in a crowd of passing people if someone displays symptoms that correspond to COVID-19. If a person displays symptoms, they will be asked by a medical person if they are willing to come to a measurement station where more symptoms can be checked. An example of a symptom that could be checked is fever, indicated by an elevated temperature.

At home, anyone can buy for example an infrared thermometer and measure their temperature. But is this also allowed at the entrance of a privately owned shopping centre? In the Netherlands, the government body responsible for the privacy of people is the Autorieit Persoonsgegevens (AP). One of the laws enforced by the AP is the Algemene Verordering Gegevensbescherming (AVG). This is a law coming from the European Union which safeguards people’s privacy. According to the AP, it is not allowed to measure someone’s temperature if the measurement data is automatically stored or processed in any way, since this is in violation of the AVG. The AVG states that it is not allowed that the measurement results in some sort of automated action, like for example a gate that opens or a light that turns green. The measurement data and the persons identity can not be stored, passed along or processed in any way. If this is followed and the person is asked permission (and is able to decline) then the AVG is not violated and measurements at the entrance of a privately owned shopping centre, like the Heuvel Galerie, are allowed. A similar procedure is already in place nearby at ASML in Veldhoven. At ASML, a temperature measurement is performed on every person who wants to enter a building. If they have a raised temperature, access is denied.

Design

In this section, first, all things taken into account during the design proces and the choices made to come to a final design are elaborated. After that the final hardware- and software design are given.

Questionnaire

A questionnaire was conducted among shoppers at the Heuvel Galerie. Here the results of the questionnaire will be discussed. Some results of this questionnaire will be incorporated in the design of the system.

RPC's

Here, the RPC's for the COVID-19 symptom detection system are given. These RPC's are based on the previously mentioned users needs.

Requirements

• Detect coughing people and notify medical staff.
• Perform IR temperature measurement on measurement station.

Preferences

• Respect privacy as much as possible.
• Easy to use for medical staff.
• Safe to use for medical staff.
• Increase the attractiveness of visiting the Heuvel Galerie.
• The system should be attractive (in terms of looks and pricing).
• No excessive hindrance for shoppers.
• Easy to install at location of use.
• Minimize amount of false positives and false negatives.

Constraints

• Comply with health and privacy regulations and laws.

Privacy

The detection system proposed in this study requires people to be taken to a measurement station when they have showed symptoms corresponding to COVID-19. The AP (Autoriteit Persoonsgegevens) warns that, when a measurement is used to for example deny someone access, and a line of waiting individuals is able to observe this happening, then this could impact the persons privacy. Measurements need to incorporated into the design to prevent this violation of privacy. The measure we chose was to also let some random people be taken to the measurement station while having a sign state that not only people with symptoms will be taken, but also random people. This way people will not think strange about a person being taken to the measurement station.

What happens after testing?

After the testing is done, the people who are tested negative can continue in the Heuvel Galerie. For people who are tested positive it gets a bit more complicated.

The Heuvel Galerie is real estate that is a part of the private company CBRE Global Investors, and any measures being taken have to go by them. It is assumed from here on forward that they agree with the measures being suggested.

When people are tested positive, it is recommended that the Heuvel Galerie denies them access. Such a temporary prohibition is legal even if it is only verbally commanded by the owner, on the grounds of the law of "huis- of lokaalvredebreuk". In article 5:1 of the "Burgerlijk Wetboek" it can be found that ownership is the most inclusive right a person can have on a company, which means that the owner decides who can go in and under which conditions. If CBRE Global Investors would write a statement that people who have been tested positive in the testconditions specified in this study are to be denied access, they have no legal right to enter the building.

By denying positively tested people acces, the Heuvel Galerie can create a safe shopping environment.

Psychological effects

In order to decrease COVID-19 spread it is very important that people are careful themselves. At the time of writing the article in September 2020, the COVID-19 pandemic has been in the Netherlands for about half a year. In the period of June 17th to June 30th, Eindhoven had 9 infected people according to the RIVM, and since then the number started exponentialy increasing to 117 infections in the period form August 26th to September 8th. Because the number of infections in Eindhoven and its regions has been relatively low for some time, the perceived risk of COVID-19 has gone down a lot. This results in people being less careful. In order to make people more aware and careful, visibility is very important. Visibility can be manifested in different ways. People you know being infected is an example of visibility. Other, more preferable examples are medi and visible coronameasures. A visible center of coronadetection of course falls under the last category. Several steps can be taken to manipulate people in being more careful by improving the design of the center and the surroundings.

People are different, and this has to be taken into account in the persuasion process. There are, on one hand, high-self-monitoring individuals, who mostly serve a social-adjustive function, and on the other hand low-self-monitoring individuals, who serve a value-expressive function. The high-self-monitoring group is more easily persuaded by messages that focus on facilitating social interaction and enhancing cohesion, and the low-self-monitoring group is more easily persuaded by messages that focus on core values or beliefs. This can be implemented in the design process by making posters for example, where one type of poster is designed to target the high-self-monitoring group, and another type of poster is designed to target the low-self-monitoring group. For the high-self-monitoring group this poster should focus on the social groups the people could be in, and imply that not adhering to the rules lowers your cohesion with the group. For the low-self-monitoring group the focus should lie on complying to the rules because of valueing for example life, health, decency and common sense.

People behave differently when they are self-aware and non-self-aware. If people are self-aware they are more likely to act according to the way they perceive themselves or how they think they should behave, which is mostly better behaviour. The presence of a mirror or an audience can greatly improve self-awareness. In case of the center for symptom detection, self-awareness of surrounding people can be increased by the presence of reflective surfaces such as mirrors and the presence of for example law enforcing people looking around.

(Next week more about perceived risk and perceived professionalism)

Questionnaire

Privacy

1. Does the procedure make you feel more protected? (1-5)

2. How much do you feel your privacy is being invaded by the symptom detection device?(1-5)

3. How much do you feel your privacy is being invaded by the procedure?(1-5)

4. What part of the procedure do you think invades your privacy most?(- being screened by the preliminary detection device - being potentially taken to a separate room for more tests - The cough test being done in the separate room -the cough test being done in the separate room -If being tested positive not being allowed to enter the Heuvel Galerie)

5. What part of the device do you think invades you privacy most? (-Infrared camera -visual camera -microphone)

6. How useful do you think the detection device will be?(1-5)

7. How useful do you think the entire procedure will be?(1-5)

8. What is you opinion about the Heuvel Galerie with the detection procedure compared to the the Heuvel Galerie without the detection procedure?(1-5)

9. Do you have any ideas to make the procedure less privacy invasive?(open)

10. Do you have any ideas to make the detection device less privacy invasive?(open)

Detection system options

The different types of detection systems that will be used in the center for COVID-19 symptom detection are thermometers for testing temperature and a camera with microphone for cough detection. The camera will not be used for detecting whether people are within a 1.5 meter range of eachother, because it is hard to distinguish between families this way. Another option that will not be implemented is fever screening through thermal imagery, because mass screening for fever requires a resolution that is so high that it does not exist yet. Screening for fever using an IR handheld thermometer could be possible, but only in a controlled environment, and it is not sensitive enough.

(Having a que or no que?)

(Coughing movement study)

Checkpoint description

Conventional Airport security Checkpoint

Design features for airport security checkpoints taken from the Transportation Security Agency (TSA) which could be applied to the symptom screening system. Some design features reported in this summary might be too oriented for security checkpoints and should be only taken for informative purpose, not as direct solutions. Very standard, very dehumanizing.

• Queue delimitation: Queue should be delimited by strap stanchions, either single or double strap. Recommended lane width is from 1m or 1.5m depending on lane function or available space. For airport use recommended queueing space is: 27m2.

• Barriers to delimit safe zone, Disability access gate: Barriers that delimit the safe zone (sterile zone) from the normal zone. Gates are placed along the delimitation to give access to people with wheelchairs or disabilities.

• Secondary screening and compose bench: The secondary screening station should be close to avoid the person to walk to much and to let the person maintain eye contact with his/ her family (1-1.5m, but now consider COVID). Mobile cabinets are used to store extra equipment. Composure bench are used for passengers to sit down and compose themselves. Benches are placed out of the passenger flow.

• Private Screening cabin: Private screening cabins are used for private pat downs of people. Private cabins should be placed close to main screening station. The recommended dimensions are 1.8x2.4x2.4m. The glass should be opaque to maintain privacy. Include power and data outlet in the design for future technology.

Standard layout:

• Queue management: For each lane there should be at least 27m2. At peak times pre-scanning or queue flexing should be used to increase throughput.

• CCTV coverage: Camera placement should avoid recording people in Private Screening Cabins. For airport security the CCTV system records and stores personal identifying data. Important design features are to maintain an accurate system time and “provide a view of people entering and exiting with enough detail to recognize them”.

• Lighting requirements: Should meet local and/or national guidelines, but 300lux (30 foot-candles). Enough lighting should be maintained the performance of the checkpoint.

Final hardware design

(CAD drawings and posters)

Software

(Pseudocodes)

Simulations

Crowds simulations:

https://github.com/trinhthanhtrung/unity-pedestrian-sim

https://unitylist.com/p/10qa/Crowd-Simulation-and-Visualization-in-Unity

https://unitylist.com/p/1qp/RVO-2-Unity

Proprietary software:

https://ucrowds.com/documentation/unity3d/manual/UUCS_Unity3D_Plugin_manual.pdf

https://arc.de/airport-challenges-in-reacting-to-covid-19/

Analysis on response to SARS-2: https://www.eurocontrol.int/sites/default/files/2020-08/covid19-impact-airport-ops-webinar-a-presentation.pdf

Cost-benefit analysis

Here, a cost-benefit analysis of the COVID-19 symptom detection system is performed.

Feasibility

Here, a conclusion is drawn on the feasibility of a COVID-19 symptom detection system.

(What are advantages compared to other COVID-19 measures, for example BOA's?)

Conclusion

Here, an overal conclusion on the project is drawn.

Conclusion

Discussion

Recommendation

Peer review

Here, the peer review is given.

Workload

An overview of who has done what per week of the project.

Week 1

Name	Student ID	Hours this week	Tasks
Davide Fabbro	1255401	9	Intro videos + first brainstorm (1.5 hours) + Studied and wrote summaries for papers: , , , , , , (6 hours) with small research on , (0.2 hours) + second meeting (0.5) + Reorganizing state of the art section (1 hour).
Lieke Nijhuis	0943276	12	Intro videos + first brainstorm (1.5 hours) + Studied and wrote summaries for papers: , , , , , , (6.5 hours) second meeting+ sending e-mail (45 minutes) + Describing problem statement and objectives (15 minutes) + Miscellaneous work (3 hours)
Ikira Wortel	1334336	10	Intro videos + first brainstorm (1.5 hours) + Studied and wrote summaries for papers , , , , , and (7 hours) + extra meeting discussing chosen subject (0.5 hour) + Who are the users (1 hour)
Wout Opdam	1241084	12.5	Intro videos + first brainstorm (1.5 hours) + Start with wiki layout (1 hour) + Studied and wrote summaries for papers: , , , and (5 hours) + wrote about approach and deliverables (0.5 hour) + Made planning for project (2.5 hours) + extra meeting discussing chosen subject (0.5 hour) + Describing user requirements (1.5 hours)

Week 2

Name	Student ID	Hours this week	Tasks
Davide Fabbro	1255401	5.5	Tutor feedback meeting (0.5 hour) + Group meeting (1 hour) + Rewrite focus on COVID-19 symptom detection (0.25 hour) + Research Thermal cameras and grid(4 hour)
Lieke Nijhuis	0943276	10	Tutor feedback meeting (0.5 hour) + Group meeting (1 hour) + Research and write "what happens after testing" (0.5 hour) + Research psychological effects detections center (5 hours) + Detection systems explanation (30 minutes)+ Extra group meeting (0.5 hour) + Miscellaneous work (2 hours)
Ikira Wortel	1334336	10	Tutor feedback meeting (0.5 hour) + Group meeting (1 hour) + Redefine USE'ers (1 hour)+ Extra group meeting (0.5 hour) + questionnaire questions privacy (2 hours) + Reading and finding papers about privacy (5 hours)
Wout Opdam	1241084	7.5	Tutor feedback meeting (0.5 hour) + Group meeting (1 hour) + Revised project planning (1.5 hours) + Re-designed wiki layout (1 hour) + Describe use case Heuvel Galerie and learn wiki page formatting (1.5 hours) + Redefine USE'er requirements (1.5 hours) + Extra group meeting (0.5 hour)

Week 3

Name	Student ID	Hours this week	Tasks
Davide Fabbro	1255401		Tutor feedback meeting (0.5 hour) + Group meeting ( 1.5 hours) + Research on similar systems (3 hours)+ Research on convetional security checkpoints (1 hour) + Miscellaneous (3 hours)
Lieke Nijhuis	0943276	12	Tutor feedback meeting (0.5 hour) + Group meeting ( 1.5 hours) + Setting up e-mail for shopowners in the Heuvelgalerie (0.5 hours) + Finding contactinfo of and mailing to 17 shopowners in the Heuvelgalerie (1 hour) + Interviewing shoppers Heuvelgalerie (2 hours) + Designing posters (3 hours) + Extra meeting (0.5 hours) + Miscellaneous work (3 hours)
Ikira Wortel	1334336	19	Tutor feedback meeting (0.5 hour) + Group meeting ( 1.5 hours) + making questionnaire (1.5 hours) + interviewing shoppers Heuvel Galerie (2 hours) + Extra meeting (0.5 hours) + Working on privacy part(8 hours) + Working on matlab model (3 hours) + Miscellaneous work (2 hour)
Wout Opdam	1241084	16	Tutor feedback meeting (0.5 hours) + Group meeting ( 1.5 hours) + Adjusting planning (1 hour) + Miscellaneous work (0.5 hours) + Research legal side of measurements at Heuvel Galerie entrance (2 hours) + Making initial design concept sketches (2 hours) + Work on Netlogo simulation model (7.5 hours) + Extra group meeting (0.5 hours) + Update user requirements and RPC's based on questionnaire (0.5 hours)

*Miscellaneous work: reading, brainstorming, discussing things in Whatsapp group etc.

References