PRE2016 3 Groep12: Difference between revisions

From Control Systems Technology Group
Jump to navigation Jump to search
 
(44 intermediate revisions by 4 users not shown)
Line 28: Line 28:
* What is the current legislation on flying drones and how can this be applied to the TU/e campus?
* What is the current legislation on flying drones and how can this be applied to the TU/e campus?
* Is it possible to change the law in case this prohibits the use of drones?
* Is it possible to change the law in case this prohibits the use of drones?
* What is the current surveillance approach at the TU/a campus?
* What is the current surveillance approach at the TU/e campus?
* How would drones improve the current TU/e security?
* How would drones improve the current TU/e security?
* How do people living on campus experience the security nowadays and how would they react to drones?
* How do people living on campus experience the security nowadays and how would they react to drones?
Line 63: Line 63:
==Product idea==
==Product idea==


We plan to use state-of-the art drones which are already equipped with cameras able to film in high resolution and framerate. We drafted a [[list of requirements]]. Afterwards we found a suitable drone candidate: DJI Phantom Pro 4.
We plan to use state-of-the art drones which are already equipped with cameras able to film in high resolution and framerate. We drafted a [[list of requirements]]. Afterwards we found a suitable drone candidate: DJI Wind 1, DJI Phantom 4 Pro, HexH20 Pro v4.
The drone is already programmable and supports autonomous flying, which means that they will follow a certain preprogrammed map, and will deliver the footage to a human operator who will be responsible for detecting criminal activity and can take control of the device at any time. This footage will be analysed directly and security teams can act based on the information. As drones are faster and more maneuverable than cameras, they are able to respond very quickly on anomalies.  
The drone is already programmable and supports autonomous flying, which means that they will follow a certain preprogrammed map, and will deliver the footage to a human operator who will be responsible for detecting criminal activity and can take control of the device at any time. This footage will be analysed directly and security teams can act based on the information. As drones are faster and more maneuverable than cameras, they are able to respond very quickly on anomalies.  
[[File:Maxresdefault.jpg|frameless|upright=4]]


===Security issues===
===Security issues===
Line 94: Line 96:
===Human detection software===
===Human detection software===


For the Human detectin software we found a framework called openCV:
For the Human detection software we found a framework called openCV:
http://opencv.org/
http://opencv.org/


We were going to try and implement this framework so that we can see from how far the drone's camera can still recognize a human, but we are no longer planning this as the focus of this project has shifted and anomaly detection is now envisioned to be done by humans instead of by the drone itself. It still could be a useful tool to draw squares around humans so that the security can easily identify them.
We are no longer planning this as the focus of this project has shifted and anomaly detection is now envisioned to be done by humans instead of by the drone itself. It still could be a useful tool to draw squares around humans so that the security can easily identify them.


===Drone charging===
===Drone charging===
Line 192: Line 194:


====Weather====
====Weather====
The next variable we take into account is the wheather. Depending on which drone we use, the drone will not always be able to fly due to wheather conditions.  
The next variable we take into account is the weather. Depending on which drone we use, the drone will not always be able to fly due to wheather conditions.  


Drones have a certain resistance to wind speed. In the case of the DJI phantom 4 this is 10 m/s wind<ref>[http://www.dji.com/phantom-4-pro/info#specs] DJI phantom 4 specs</ref>. When looking at wind data gathered over a time period of about 30 years, we can conclude that this drone can fly in 98.12% of all time because of the wind<ref>[http://projects.knmi.nl/hydra/cgi-bin/freqtab.cgi?language=nl] KNMI wind statistics</ref>. Our tool will take into account wind resistance depending on which drone will be used.
Drones have a certain resistance to wind speed. In the case of the DJI phantom 4 this is 10 m/s wind<ref>[http://www.dji.com/phantom-4-pro/info#specs] DJI phantom 4 specs</ref>. When looking at wind data gathered over a time period of about 30 years, we can conclude that this drone can fly in 98.12% of all time because of the wind<ref>[http://projects.knmi.nl/hydra/cgi-bin/freqtab.cgi?language=nl] KNMI wind statistics</ref>. Our tool will take into account wind resistance depending on which drone will be used.
Line 201: Line 203:
After a meeting with the head of TU/e security (interview below) we found that in general cameras will be replaced in cycles of about 10 years. Therefore we decided to calculate the costs of introducing our technology over a period of 10 years.  
After a meeting with the head of TU/e security (interview below) we found that in general cameras will be replaced in cycles of about 10 years. Therefore we decided to calculate the costs of introducing our technology over a period of 10 years.  


The biggest part of the cost will be the amount of drones to be used. Each drone has a certain expected lifetime, the cameras were replaced after 10 years but for a drone the expected lifetime is between 2 - 6 years. The next cost we take into consideration is the cost of the spare batteries we use in order to keep the drone up in the air as much as possible. Next we consider the costs of power consumption, the drone uses more power than the regular cameras since they have te stay up in the air all the time. The energy cost is computed as follows:
The biggest part of the cost will be the amount of drones to be used. Each drone has a certain expected lifetime, the cameras were replaced after 10 years but for a drone the expected lifetime is between 2 - 6 years. The next cost we take into consideration is the cost of the spare batteries we use in order to keep the drone up in the air as much as possible. Next we consider the costs of power consumption, the drone uses more power than the regular cameras since they have te stay up in the air all the time. The energy cost for the drones is computed as follows:


energycost = uptime*energy/(1000*batterylife)*87658.2/Kwhcost
energycostDrones = uptime*energy/(1000*batterylife)*87658.2/Kwhcost


Here energy denotes the amount of watt hours that a single battery has and 87658.2 is the amount of hours in 10 years. Finally we also consider software costs. In order to let the drones fly autonomously we need to program the coordinates between which the drones will fly. We also need the security personnel to be able to stop this autonomous flying and manually look at certain places when they see something suspicious. Finally the drones should communicate the video footage through a secure channel. This function does not ship with most drones, so it will have to be programmed in.
Here energy denotes the amount of watt hours that a single battery has and 87658.2 is the amount of hours in 10 years. Finally we also consider software costs. In order to let the drones fly autonomously we need to program the coordinates between which the drones will fly. We also need the security personnel to be able to stop this autonomous flying and manually look at certain places when they see something suspicious. Finally the drones should communicate the video footage through a secure channel. This function does not ship with most drones, so it will have to be programmed in.
Line 212: Line 214:


cost = dronecost*droneAmount*10/lifetime
cost = dronecost*droneAmount*10/lifetime
+ batterycost*Bamount*droneAmount + energycost - cameracost*cameraAmountReplaced - 10*YearlySalary*EmployeesReplaced
+ batterycost*Bamount*droneAmount + energycostDrones - cameraAmountReplaced*energyCostCamera - cameracost*cameraAmountReplaced - 10*YearlySalary*EmployeesReplaced


====The tool====
====The tool====
Github: [https://github.com/jamiro24/FinancialModel/]
Runnable jar file: [https://mega.nz/#!TsYAXKgA!4hTH1qMcHfGNL67zwqL0mBpQvM4GVQd3EbUTY1t-bKU]
[[File:tool.PNG|frameless|upright=4]]
[[File:tool.PNG|frameless|upright=4]]


This section will describe the tool that was produced in order to do the calculations discussed in previous sections. The image above shows the graphical user interface. The "drone preset" option can be used to select different drones. This is useful since different drones will result in a different cost in the end. New drones can be added by pressing the "+" sign. The next option is used in order to define how big the area is that needs to be surveilled. In the case of the TU/e campus this is 0.74 km^2. The "view distance" option is used to define how far the drone should be able to look. Since currently the cameras at the TU/e campus can look about 100 - 150 meters we defined the default value to be 150 meters.  
This section will describe the tool that was produced in order to do the calculations discussed in previous sections. The image above shows the graphical user interface. The "drone preset" option can be used to select different drones. This is useful since different drones will result in a different cost in the end. New drones can be added by pressing the "+" sign. The next option is used in order to define how big the area is that needs to be surveilled. In the case of the TU/e campus this is 0.74 km^2. The "view distance" option is used to define how far the drone should be able to look. Since currently the cameras at the TU/e campus can look about 100 - 150 meters we defined the default value to be 150 meters.  


The next option is the average distance to the charge station. In a previous section we have shown how we got the value of 0.4 km for the TU/e. The calculations will be based on a single charging station which will be enough for areas like the TU/e. The "height" option defines how high the drone will fly. In general the drone will be able to see more at once when flying at a higher heights, however as will be discussed later on there are restrictions with respect to law and also on the distance at which people can clearly be distinguished. The "speed" option defines the default speed at which the drone will fly and the "time frame" option will define in what time the drones should be able to cover the entire campus. Based on this time frame the amount of drones needed will be calculated. The "tilt angle" corresponds to the alpha angle from the computations above and defines the angle at which the camera is located.  
The next option is the average distance to the charge station. In a previous section we have shown how we got the value of 0.4 km for the TU/e. The calculations will be based on a single charging station which will be enough for areas like the TU/e. The "height" option defines how high the drone will fly. In general the drone will be able to see more at once when flying at a higher heights, however as will be discussed later on there are restrictions with respect to law and also on the distance at which people can clearly be distinguished. The "speed" option defines the default speed at which the drone will fly and the "time frame" option will define in what time the drones should be able to cover the entire campus. Based on this time frame the amount of drones needed will be calculated. The "tilt angle" corresponds to the alpha angle from the computations above and defines the angle at which the camera is looking at the ground.  


Finally some text boxes are in place that can be used to define some cost values. In the case of the TU/e the cameras cost about 3500 euros based on the interview with the head of security. The next option defines how many cameras will be replaced. On the TU/e campus there are 50 cameras in total. Next we have options to define the yearly salary of an employee and how many should be replaced. Right now 3 employees are always in use. The last option is to define the software cost that comes with programming the drones.
Finally some text boxes are in place that can be used to define some cost values. In the case of the TU/e the cameras cost about 3500 euros based on the interview with the head of security. The next option defines how many cameras will be replaced. On the TU/e campus there are about 50 cameras in total. Next we have options to define the yearly salary of an employee and how many should be replaced. Right now there are always 3 security employees at the TU/e, each with a yearly salary of about 50000 euros. The last option is to define the software cost that comes with programming the drones.


To the right of these settings the output of the calculations can be seen. At the top some values for the particular drone used are shown. These values include drone cost, battery life and field of view of the camera for example. The next value, the coverage, is the percentage of the campus that is seen by the drone within the timeframe. In this case this is 20.99% of the campus. From this percentage we calculate that we need 5 drones to see the entire campus in 15 minutes. The next value shown is the uptime, based on battery life and wheather effects. Here the drone not being waterproof has the biggest impact. The next couple of lines display different costs. The main thing to look at is the "Total additional costs for 10 years" line. This line says how much more expensive (negative value means cheaper) it is to deploy the drones. In this case it costs 41065.07 euros more than the current situation.
To the right of these settings the output of the calculations can be seen. At the top some values for the particular drone used are shown. These values include drone cost, battery life and field of view of the camera for example. The next value, the coverage, is the percentage of the campus that is seen by the drone within the timeframe. In this case this is 20.99% of the campus. From this percentage we calculate that we need 5 drones to see the entire campus in 15 minutes. The next value shown is the uptime, which is based on battery life and wheather effects. Here the drone not being waterproof has the biggest impact. The next couple of lines display different costs. The main thing to look at is the "Total additional costs for 10 years" line. This line says how much more expensive (negative value means cheaper) it is to deploy the drones. In this case it costs 41065.07 euros more than the current situation.


We can clearly see that currently the salary of the security employees is the biggest part of the cost. If we would be able to replace an employee we would be able to deploy the drones without any issue in this situation. However it may not be feasible to replace an employee since there are only three (at the same time).
We can clearly see that currently the salary of the security employees is the biggest part of the cost. If we would be able to replace an employee we would be able to deploy the drones without any issue in this situation. However it may not be feasible to replace an employee since there are only three (at the same time).


===Conclusion===
===Conclusion===
Using this tool we can find certain break even points with respect to the cost. When using the DJI phantom 4 pro drone at a height of 25 meters, with a speed of 10 km/h, a camera tilt angle of 30 degrees and a total coverage duration of 15 minutes we can find that we need to replace 13.571734285714 cameras in order to have 0 extra costs over a period of 10 years. We consider these settings to be close to the actual settings that will be used eventually, changing the height or the angle only changes the cost slightly. The main thing to test experimentally is at which speed the video feed of the drones can still be adequately checked by employees, our tool obviously concludes that the higher the speed of the drones the less drones we will need. Replacing about 14 cameras should not be a problem at all since there are 50 at the moment and the drones will be able to see everything in 15 minutes making most cameras unnecessary. The uptime of this drone (40.87%) is not that high due to the wheather effects discussed above. The DJI Wind on the other hand has an uptime of 99.33% because of it's great wind and rain resistance. With this big uptime we would be able to replace way more of the cameras. If we were to use the same settings as with the previous calculation and replace all 50 cameras we find that there is an additional cost of only 51155.69 euros. This is not much at all considering that this is over a period of 10 years. If we take it to the extreme, where we fly at 30 meters and have a camera tilt of 45 degrees we see that we even save 23778.41 euros with respect to the current situation. We see that from a financial point of view we do not necessarily have to replace seucrity employees, since not much cost is introduced. Again experiments will have to show to what extend drones can replace security personnel, but for now it seems that replacing will not be necessary.  
Using this tool we can find certain break even points with respect to the cost. When using the DJI phantom 4 pro drone at a height of 25 meters, with a speed of 10 km/h, a camera tilt angle of 30 degrees and a total coverage duration of 15 minutes we can find that we need to replace 34 cameras in order to have no extra costs over a period of 10 years. We consider these settings to be close to the actual settings that will be used eventually, changing the height or the angle can have a big impact on the cost, but flying to high or to low or looking at the ground at a high angle will make the system harder if not impossible to use, as you will not be able to see what is going on. We think this height and angle are balanced, such that we can cover a lot of ground, but don't lose functionality. The main thing to test experimentally is at which speed the video feed of the drones can still be adequately checked by employees, our tool obviously concludes that the higher the speed of the drones the less drones we will need. Replacing about 34 cameras can be a problem when using the phantom 4 since the uptime of this drone is only 40.87%. This would mean that during the other 59.13% of the time we would still have to rely on the cameras. When we replace 34 out of the 50 we will lose too much coverage of the campus.The low uptime of this drone (40.87%) is not that high due to the wheather effects discussed above. The DJI Wind on the other hand has an uptime of 99.33% because of it's great wind and rain resistance, when we want to replace cameras we need this greater uptime. If we were to use the same settings as with the previous calculation and replace all 50 cameras we find that there is an additional cost of 276004.87 euros. If we take it to the extreme, where we fly at 30 meters and have a camera tilt of 45 degrees we see that we still have an additional cost of 125669.91 euros with respect to the current situation (Though in this situation the drones might not be as effective due to the high height and large angle). This big increase in cost can be explained by the bigger uptime, since when we fly all the time we will use more energy which will cost more money. The DJI Wind 1 itself also costs more money, 10 times more than the phantom 4. However this is because the Wind 1 is a custom made drone, in the future when more wheather resitant drone will come on the market this increase rate will go down. In the first case of the DJI Wind 1 we see that both the energy cost and the drone cost amount for about 50% of the total costs. We found a drone, the HexH20 pro V2, that is cheaper in cost. This one only costs 4 times as much as the phantom 4, however it uses way more power. Using the same settings we see that we have an additional cost of 675385.86 euros, which is more than the extra 276004.87 that we got for the Wind 1.


In the calculations above we left the software costs out, since this cost is the hardest to estimate.
In the calculations above we have not considered yet to replace employees, if we could replace a single employee we would be able to introduce the Wind 1 drones without any issue and still save about 200000 over a period of 10 years. However we should test experimentally to what extend we can replace employees, since this might be difficult as there are only 3 at the moment. 
 
In the calculations above we left the software costs out, since this cost is the hardest to estimate. This cost is usually underestimated and can be rather big. We estimate this cost to be between 10000 - 150000 euros, meaning that we will definetely have to replace at least a single employee if we want to introduce the technology without any extra cost.  With the remaining money we can keep at least 10 cameras, meaning that we still have some reliable reference points of the most important areas of the campus.
 
he found break even points are not really a realistic setting as the TU/e security was not open to replacing employees and replacing almost all cameras is probably not a good idea either, since the largest uptime for the drones we used was 91.29%, so that would mean that the TU/e security would be blind for about 10% of the time, which is unacceptable. This is why we've done a calculation for a situation we think is realistic for each drone type. We used the following settings for this calculation:
* Height: 25m
* Speed: 10 km/h
* Tilt angle: 30 degrees
* Amount of cameras replaced: 20
* Amount of employees replaced: 0
* Software costs: 150000
We chose these settings, such that we don't fly to high and lose visibility, we don't fly to fast and get blurry video footage, we don't have a too large angle so that we can still see what's below the drone and we keep enough cameras to not be blind when the drones can't fly. Using these settings we got the following additional costs:
* DJI Phantom 4 Pro: 200882.08 euros
* DJI Wind 1: 542575.75 euros
* HexH2O Pro V2: 941953.87 euros
While the DJI Phantom 4 Pro is the cheapest option, it can only fly 40.87% of the time, due to not being waterproof. We don't think this would be a viable option for securing the TU/e, So we would rather go for the DJI Wind 1, which is the second cheapest option. Still this seems to cost too much for the TU/e security.
Even if this drone becomes cheaper in the future (it costs 14950 euros now) It would still be too expensive due to the costs of electricity. Lets say it will cost 5000 euros in the future, the additional cost will still be about 300000 euros, which is probably still too expensive.
So to conclude, financially it is not feasible to implement drones for security at the TU/e, if we don't replace any employees. If the TU/e security would agree to replace 1 employee for the drones, it would be feasible, but the TU/e security really didn't like this idea, so we don't think that will ever happen.


==Law analysis==
==Law analysis==
Line 375: Line 404:
In order for our idea to work these rules have to be removed, or an exemption has to be made for certain organisations. This seems unlikely though, as the prototype regulation presented by the European Commission also contains these rules.
In order for our idea to work these rules have to be removed, or an exemption has to be made for certain organisations. This seems unlikely though, as the prototype regulation presented by the European Commission also contains these rules.


==Survey analysis==
==Societal Analysis ==
The implementation of drones in daily, security TU/e campus life will be a social matter as well as a technological one. We decided to analyze three visions on safety guarantee by drone implementation, namely the perspective of:
The implementation of drones in daily, security TU/e campus life will be a social matter as well as a technological one. We decided to analyze three visions on safety guarantee by drone implementation, namely the perspective of:


Line 469: Line 498:


===Survey===
===Survey===
We are going to deploy this [https://docs.google.com/forms/d/e/1FAIpQLSd243OSchI_cXfztvwbiIQZpOIlAJqJWnqsXXkbIjaML2I3Fg/viewform?usp=sf_link survey] to the students living on the campus (main goal) and studying in the TU/e (secondary goal).
We deployed this [https://docs.google.com/forms/d/e/1FAIpQLSd243OSchI_cXfztvwbiIQZpOIlAJqJWnqsXXkbIjaML2I3Fg/viewform?usp=sf_link survey] to the students living on the campus (main goal) and studying in the TU/e (secondary goal).


===Analysis of results===
===Analysis of results===
Line 482: Line 511:
1.The first question we asked the students is: “Do you think that TU/e campus provides a safe environment for students?”. Here are the results:
1.The first question we asked the students is: “Do you think that TU/e campus provides a safe environment for students?”. Here are the results:


[[File:Screenshot_1.png]]
[[File:Screenshot_1.png|frameless|upright=4]]


Students in general think the environment on the TU/e is very safe, as you see above students could rate how safe they feel on the TU/e from 1 to 4. Where 1 is considered very bad and 4 is very good. Most students picked 4 and feel very safe on campus, the rest all picked 3. We can conclude from this that people feel safe with the current security.
Students in general think the environment on the TU/e is very safe, as you see above students could rate how safe they feel on the TU/e from 1 to 4. Where 1 is considered very bad and 4 is very good. Most students picked 4 and feel very safe on campus, the rest all picked 3. We can conclude from this that people feel safe with the current security.
Line 497: Line 526:
In the graph below you see the results of the question: Do you think security should be improved? 68.6% of people answered no here, which means most people think security as fine as it is. Still almost 3 out of 10 people have other thoughts about this, 22.9% say Yes here and 8.6% say “Other”, which were in general: It’s good as it is but it always can be improved, and 1 perhaps.
In the graph below you see the results of the question: Do you think security should be improved? 68.6% of people answered no here, which means most people think security as fine as it is. Still almost 3 out of 10 people have other thoughts about this, 22.9% say Yes here and 8.6% say “Other”, which were in general: It’s good as it is but it always can be improved, and 1 perhaps.


[[File:Screenshot_2.png]]
[[File:Screenshot_2.png|frameless|upright=3]]


5. The fifth question we asked was: “How do you think the security be influenced, if we substitute some camera's by surveillance drones?”. Here are the results:
5. The fifth question we asked was: “How do you think the security be influenced, if we substitute some camera's by surveillance drones?”. Here are the results:


[[File:Screenshot_3.png]]
[[File:Screenshot_3.png|frameless|upright=4]]




Line 508: Line 537:
6. The sixth question was: “Would you mind if there would be autonomous drones flying around campus at all times?”. Here are the answers:  
6. The sixth question was: “Would you mind if there would be autonomous drones flying around campus at all times?”. Here are the answers:  


[[File:Screenshot_4.png]]
[[File:Screenshot_4.png|frameless|upright=3]]


By “other”, people mostly said that they might mind in the beginning or were not sure about it.  
By “other”, people mostly said that they might mind in the beginning or were not sure about it.  
Line 514: Line 543:
7. How much do you like this drone idea?
7. How much do you like this drone idea?


[[File:Screenshot_5.png]]
[[File:Screenshot_5.png|frameless|upright=4]]


In this question we asked how they liked our idea. The students could rate this from 1 (Don’t like it at all) to 6 (I really like this idea). The results were really diverse and it seems people mostly have a strong opinion about this, either they really liked it or they really didn’t. We were surprised to see that so many people didn’t like the idea of security drones.
In this question we asked how they liked our idea. The students could rate this from 1 (Don’t like it at all) to 6 (I really like this idea). The results were really diverse and it seems people mostly have a strong opinion about this, either they really liked it or they really didn’t. We were surprised to see that so many people didn’t like the idea of security drones.


===Conclusion===
===End results===
What we can conclude from this is that people are very satisfied with the security at this moment. When we asked them about deploying drones the results were very diverse. We were surprised that so many students were pessimistic about drone security. There were more students which didn’t like the idea (57%) compared to the students that did like the idea (43%). Maybe in the future, when drones become more familiar to the public and do not disturb them as much, people would accept them easier.
What we can conclude from this is that people are very satisfied with the security at this moment. When we asked them about deploying drones the results were very diverse. We were surprised that so many students were pessimistic about drone security. There were more students which didn’t like the idea (57%) compared to the students that did like the idea (43%). Maybe in the future, when drones become more familiar to the public and do not disturb them as much, people would accept them easier.


==Planning==
==Conclusion==
 
Gantt chart:
https://drive.google.com/open?id=1bb1QMkg4TYDTVDhT7nHERBCg_1VZtb98BMwWiS9wT6I


==Conclusion==
By means of conclusion, looking back at our objective questions,  at this point in time it is not financially feasible to replace cameras and add drones for surveillance on campus. We derived those results by utilizing our financial model which we created with respect to all relevant laws and rules concerning drones. We found that filming people on campus is not going to be an issue for TU/e security, since they have a licence for filming anyway. Despite that, flying drones autonomously is an issue because current legislation states that a person should always have the device at a line-of-sight distance. In addition, we conducted interviews with the head of security of the Technical University on one side and a police officer specialized in drone authorisation on the other. Finally, we found out that most people are sceptical of the idea by conducting surveys. Society is not at that point of familiarity with drones to let them fly autonomously and film their every move.


==Discussion==
==Discussion==
Line 535: Line 561:
Moreover, the difference in area surveilled by a drone moving in a straight line compared to moving in other directions (dodging obsticles, change of direction, etc.) is measured by a variable which we do not know if it is accurate right now. The only way of knowing that is by experimenting with actual devices.
Moreover, the difference in area surveilled by a drone moving in a straight line compared to moving in other directions (dodging obsticles, change of direction, etc.) is measured by a variable which we do not know if it is accurate right now. The only way of knowing that is by experimenting with actual devices.
Another point of improvement is to take into account the ability of operators to detect anomalies using moving cameras (drones) compared to static cameras. Also at what survellance drone speed does the anomaly detection abilities of the person watching the video hinder.  
Another point of improvement is to take into account the ability of operators to detect anomalies using moving cameras (drones) compared to static cameras. Also at what survellance drone speed does the anomaly detection abilities of the person watching the video hinder.  
Besides, we could take into account the impact of future technology development. Price of technology decreases so drone options will be more realistic. This transitions can be included in the model.


Survey:
Survey:
We tried to keep the survey as short as possible because we thougth that people would not fill it otherwise, but as a point of improvement we could have included a detailed explanation of the idea and why it is good. The lack of such details might be the reason why a large part of the people who took the survey were not a fan of the concept.
We tried to keep the survey as short as possible because we thougth that people would not fill it otherwise, but as a point of improvement we could have included a detailed explanation of the idea and why it is good. The lack of such details might be the reason why a large part of the people who took the survey were not a fan of the concept.


Line 587: Line 615:


'''Week 7'''
'''Week 7'''
''Weekly feedback session between individual groups and expert panel. Feedback is on both learning goals (technical content and process).''
During this week we made the survey and asked hundress of students (most living on campus) about their opinion. We got 36 replies and analysed the results.
 
Furthermore, we finalized the business model and worked on some bug fixes.
 
Also we did work on this wiki, adding and changing different sections (for example: made our objectives clearer and made sure we answered all the needed questions).


'''Week 8'''
'''Week 8'''

Latest revision as of 11:15, 3 April 2017

Group members

  • Bibi Huijgen - 0906203
  • Jamiro Leander 0890417
  • Jari de Kroon - 0888168
  • Juan van der Heijden - 0898805
  • Plamen Pasliev - 0890518
  • Stan Roelofs - 0892914

Introduction

University of Technology Eindhoven has been a place of studying and innovation since 1986. The last decades, the board has decided to add the opportunity of living to the campus. The great amount of students functioning on a daily basis added up to the students living at the TU/e, result in a very large total of people on campus grounds. And where people are, is crime. In 2015, around 57% of all crimes recorded, was related to theft and burglary. Of course, the TU/e is currently protected: we see surveillance all the time and cameras are obvious. However, as we are educated to see technological opportunities, the TU/e security can be improved.

A very futuristic and technologically advanced solution to theft and burglary on the campus, is the use of drones. This technology allows a reduction in manpower and a more flexible way of surveillance compared to static cameras, too. By flying around the campus, theft and burglary can be detected very soon and this could reduce crime rates at the TU/e. Before the implementation of a drone system can be realized, a lot of research has to be done.

First, the law can be a restricting factor. Privacy and no-flyzones are important issues when making use of drones. Second, the drones have to be safe. They fly above a lot of people everyday and injuries cannot be risked. Besides, the technology must be advanced enough to cover the whole campus under any condition. Finally, the use of drones will be no option if their costs become too high. Obviously, the financial aspects are of great importance.

An other important part influencing the need of security improvement, is the experienced safety by TU/e students. Their perception of security will alarm or ease the current reps.

Implementing drones as a part of the TU/e security system is a challenging research topic. The rate of crimes could even be decreased as this technology will allow for faster response times to other related crimes and hazards. The overall goal is to give the citizens an increased sense of security and safety: improving the security system by using drones, taking into account the factors of law, technological possibilities and money.

Objectives

An important part of research, is formulating objectives and milestones. The objectives mentioned below, will all be supporting one main research question. This research question guides the research and gives way to new sub questions. Our main research question is:

How can drones be used as part of the TU/e campus security team and what is their effect on the surveillance costs?

To answer this question, a few minor questions have to be answered.

  • What is the current legislation on flying drones and how can this be applied to the TU/e campus?
  • Is it possible to change the law in case this prohibits the use of drones?
  • What is the current surveillance approach at the TU/e campus?
  • How would drones improve the current TU/e security?
  • How do people living on campus experience the security nowadays and how would they react to drones?
  • What is the influence of drones on financial aspects?

These questioned can be answered by setting the following objectives.

  • Research what the effects on society will be when security drones are deployed.
  • Research what the financial benefit would be to deploy surveillance drones.
  • Develop a relevant model to show the financial impact of drone use.
  • Research the legal issues regarding surveillance drones.
  • Find out what the police thinks about surveillance drones.
  • Find out what TU/e campus inhabitants think about surveillance drones

Deliverables

When working on this project, the needed deliverables have to be kept in mind. To create an underlying structure, the following interconnected deliverable are needed:

  • Survey results that show the opinion of the TU/e campus inhabitants on the use of drones
  • Interview with people responsible for the TU/e security
  • Interview with a police officer to invest what the position of governmental security is
  • Analysis of the current law and how it may be adjusted
  • Analysis and conclusion of the surveys and interviews
  • Model that shows financial impact of several drone related changes to the security system
  • Weekly modified Wiki page to keep track of the progression and create a report

After completing all steps, the conclusion and process will be shown in a short presentation.

Approach

  • Research state of the art technologies using books and articles.
  • Make an analysis of the current legislation regarding relevant topics such as filming humans and flying drones. Find out how privacy can be preserved and finally investigate if the current laws can be changed and how.
  • Find out what people think of drones surveilling their neighbourhoods by making a survey and interviewing people.
  • Make a financial model in Java where the input is an area and level of security and the output is the number of drones, cameras, or people that are needed to watch the area. To make the model we first have to use a drone to find out how high a drone can fly while still detecting humans without issues. We also need to know the costs of drones, security cameras, and people, and how long drones can fly before they have to be charged.

Product idea

We plan to use state-of-the art drones which are already equipped with cameras able to film in high resolution and framerate. We drafted a list of requirements. Afterwards we found a suitable drone candidate: DJI Wind 1, DJI Phantom 4 Pro, HexH20 Pro v4. The drone is already programmable and supports autonomous flying, which means that they will follow a certain preprogrammed map, and will deliver the footage to a human operator who will be responsible for detecting criminal activity and can take control of the device at any time. This footage will be analysed directly and security teams can act based on the information. As drones are faster and more maneuverable than cameras, they are able to respond very quickly on anomalies.

Maxresdefault.jpg

Security issues

To gain insight on this level, contact was made with the police office of Eindhoven and the head of security of the TU/e campus. The information from these sources is of utmost importance to the development the drone implementation. Besides the interview, laws and legislation is provided on the World Wide Web.

Stakeholders

There are several parties involved and every party has its own particular position. We will define how all are involved:

Government

Police Office Eindhoven

Eindhoven municipality

TU/e security

TU/e Executive Board

Campus citizens and visitors

Scenario

The scenario in which we will investigate the application of drones for security is at the campus of the TU/e. The aim is to replace (some) cameras that are located outside, giving security personnel more freedom in viewing the campus. Besides this increase in surveillance possibitlities, we will investigate the financial and legal side of this technology as well.

In this scenario drones will fly around campus autonomously by default, however security personnel should be able to take over the controls in order to double check a possible security threat for example. Security personnel will view the live feed of the drones. Since drones are moving it can sometimes be a bit more difficult to see some things. In order to help the personnel with this some post-processing on the videofeed can be done, for example drawing squares around humans using human detection software. This way the personnel can quickly see where to look on the video feed. A recharge station will be located somewhere in the center of the campus, drones will fly towards this station once their batteries are below a certain threshold. In the case of TU/e campus a single recharge station seems reasonable, however if this technology would be applied for larger areas multiple stations may be needed. At the recharge station the battery of the drone will be replaced such that the drone can immediately go back to its surveillance routine. The empty battery will be recharged and can be used for other drones once recharged.

Financial Model

The financial model will be a tool that can be used in order to derive conclusions on the cost effectiveness of applying drones. It will take into account many variables such as wheather, hardware costs of the drones as well as the cost of cameras that are currently used. All of this will be combined into one graphical user interface (GUI) where variables can be assigned values and results can be interpreted. There will be options for manual calculations, as well as options where the tool will optimize certain values, for example the number of drones to be used. For this project we will take the TU/e campus as an example where drones will be applied, however the tool will be developed in such a way that it can be applied to different areas as well.

Human detection software

For the Human detection software we found a framework called openCV: http://opencv.org/

We are no longer planning this as the focus of this project has shifted and anomaly detection is now envisioned to be done by humans instead of by the drone itself. It still could be a useful tool to draw squares around humans so that the security can easily identify them.

Drone charging

Costs

Listed below are the factors we think will play a role in the costs of implementing this system in society:

Drones

One time costs

  • Cost of a drone
  • Cost of camera
  • Cost of charging stations
  • Amount of drones needed
  • Amount of charging stations needed

Continues costs

  • Range drone can search (height detection, speed)
  • Battery life
  • Cost of electricity
  • Lifetime of a drone
  • Detection efficiency

Cameras

  • Cost of cameras
  • Lifetime of a camera
  • Amount of cameras needed
  • Percentage of ground covered
  • Detection efficiency

Police

  • Police salary
  • Cost of police car
  • Amount of police needed
  • Amount of police cars needed
  • Covered ground
  • Detection efficiency

Model

Area

First of all the area where the drones will fly should be defined. In our case this will be the area of the TU/e campus. An external tool [1] was used to find this area which is equal to 0.74 km^2, this is shown in the image below.

TueArea.PNG

Besides the size of the area to be scanned, the size a drone can see at once is also of importance. Because of the way how cameras work, this area is a trapezoid. In order to calculate this area, we define the front view of the drone to be the side where the camera is located and we define the side view of the drone to be on either side of this front view. Furthermore we define the following variables:

  • h = the height the drone is flying
  • alpha = the tilt angle of the camera of the drone
  • 2*beta = the vertical field of view of the camera in degrees
  • 2*phi = the horizontal field of view of the camera in degrees

2*phi depends on the camera that is used for the drone. For the DJI pahtom 4 pro for example this angle is 94 degrees. However since we want to use this financial model with different types of drones we decided to keep this as a variable. The images below describe the area which the drone can see. The image on the left describes the situation from the side view, the other image descirbes the situation from the front/back view.

Viewangle.png Width.png

Then, assuming we never look past the horizon, the area A the drone can see is:

  • A = .5*(wclose+wfar)*d[2], where:
  • wfar = 2*dfar*tan(phi)
  • wclose = 2*dclose*tan(phi)
  • dfar = h/cos(beta+alpha)
  • dclose = h/cos(beta-alpha)
  • d = h*tan(beta+alpha) + h*tan(beta-alpha)

We can now already calculate some simplified statistics. If we would ignore all other factors, we can calculate how much of the area of the TU/e campus can be covered by the drone given its speed, height, tilt angle of the camera and the time frame t. One last thing that we took into consideration is that you can't see as good at a distance as you can up close, so if dfar is longer than a 150 meters, we cut it off at 150 meters and recalculate beta for all 'far' variables. This cutting of the distance can be seen in the image below, where dfar exceeds 150 meters and a new beta is calculated using the new distance dfar new (150 meters).

Dfarcutting.png

For example for a speed of 10 km/h, a height of 25 meters, a camera tilt of 25 degrees and a time frame of 15 minutes we get the following result:

Dronestats.PNG

Here surfaceArea corresponds to the area A that the drone covers at any point, widthFar corresponds to wFar, move Area corresponds to the total area that the drone has seen. Finally the percentage of the TU/e campus that is covered is shown. In this case 32.88 percent of the campus was covered in 15 minutes. If we were to use three drones then the entire campus would be covered in 15 minutes.

This calculation however is missing some important things still. First of all drones cannot fly all the time due to battery life and the wheather effects to some extend. In the following sections we will discuss how we will deal with these aspects in our model.

Battery Life

The drone will not be able to fly forever of course. The drone has a certain battery life, after a certain amount of time the battery will have to be recharged. In order to achieve this we choose for having multiple batteries and replacing the empty one with a charged one such that the drone can immediately continue with its surveillance route. Replacing the battery takes a certain amount of time, first the drone needs to fly to the recharge station. It can do this at a higher speed than when flying for the surveillance.

The drone will not always be at the same distance from the recharging station. We will take the average distance to the recharging station for the calculation here. Consider the following image: Tucircle.PNG

The charging station should be in a central location. The image above uses the Laplace building as a central location, which is also where the TU/e security is located. The circle has a range of 600 meters. The average distance (Davg) to the center of the circle is 2/3 * 600 is 400 meters[3]. We assume that the drone will fly at full speed (Vmax) back towards the charging station. The final thing that we need to know is how long it will take for the drone to land, the battery to be replaced and the drone to take off again. The landing time (Treload) depends on the height the drone flies of course, however it will not take very long. We estimate this value to be between 3 to 5 minutes, this estimation was received by a user of the DJI Phantom 3 camera drone. We can now calculate the uptime (UPb) of a drone with respect to its battery life. This value is equal to:

UPb = batterylife / (batterylife + Treload + Davg / Vmax)

The amount of batteries we will need depends on how long the batteries will need te fully recharge and how long the drone can stay up in the air with a single battery. The amount of batteries (Bamount) is defined as:

Bamount = Ceil[ 1 + (charge time)/(batterylife) ]

For example when the battery needs 20 minutes before it is fully charged and the batterylife is 10 minutes, then we will need 3 batteries. One will be in the drone, and the two remaining batteries will be enough to always have one fully charged to replace the drone's battery with.

Weather

The next variable we take into account is the weather. Depending on which drone we use, the drone will not always be able to fly due to wheather conditions.

Drones have a certain resistance to wind speed. In the case of the DJI phantom 4 this is 10 m/s wind[4]. When looking at wind data gathered over a time period of about 30 years, we can conclude that this drone can fly in 98.12% of all time because of the wind[5]. Our tool will take into account wind resistance depending on which drone will be used.

Besides wind there is also a lot of rain in the Netherlands. Most drones cannot fly in the rain since they are not waterproof. A waterproof drone will cost more money, but this might be a smart thing to invest in. In the Netherlands there is rain every other day when looking at statistics from 1981-2010[6]. In Eindhoven it will rain in about 170 out of 365 days, meaning that we will not be able to fly the drones fully in those days. We can now calculate a total uptime of a drone by multipling the UPb with both the wind and rain percentages. We will use this number later on for calculating the cost of the power consumption of each drone, since when we are not flying we do not use power.

Calculating costs

After a meeting with the head of TU/e security (interview below) we found that in general cameras will be replaced in cycles of about 10 years. Therefore we decided to calculate the costs of introducing our technology over a period of 10 years.

The biggest part of the cost will be the amount of drones to be used. Each drone has a certain expected lifetime, the cameras were replaced after 10 years but for a drone the expected lifetime is between 2 - 6 years. The next cost we take into consideration is the cost of the spare batteries we use in order to keep the drone up in the air as much as possible. Next we consider the costs of power consumption, the drone uses more power than the regular cameras since they have te stay up in the air all the time. The energy cost for the drones is computed as follows:

energycostDrones = uptime*energy/(1000*batterylife)*87658.2/Kwhcost

Here energy denotes the amount of watt hours that a single battery has and 87658.2 is the amount of hours in 10 years. Finally we also consider software costs. In order to let the drones fly autonomously we need to program the coordinates between which the drones will fly. We also need the security personnel to be able to stop this autonomous flying and manually look at certain places when they see something suspicious. Finally the drones should communicate the video footage through a secure channel. This function does not ship with most drones, so it will have to be programmed in.

We also save on some costs. When we replace some of the cameras by the drones we will save on these costs. We could also save on costs by replacing personnel. For each replaced employee we save 10 yearly salaries, since we look at a period of 10 years.

The total additional cost for 10 years (to get the total cost we add what the tue currently spends on employees and cameras) will be calculated by:

cost = dronecost*droneAmount*10/lifetime + batterycost*Bamount*droneAmount + energycostDrones - cameraAmountReplaced*energyCostCamera - cameracost*cameraAmountReplaced - 10*YearlySalary*EmployeesReplaced

The tool

Github: [15]

Runnable jar file: [16]

Tool.PNG

This section will describe the tool that was produced in order to do the calculations discussed in previous sections. The image above shows the graphical user interface. The "drone preset" option can be used to select different drones. This is useful since different drones will result in a different cost in the end. New drones can be added by pressing the "+" sign. The next option is used in order to define how big the area is that needs to be surveilled. In the case of the TU/e campus this is 0.74 km^2. The "view distance" option is used to define how far the drone should be able to look. Since currently the cameras at the TU/e campus can look about 100 - 150 meters we defined the default value to be 150 meters.

The next option is the average distance to the charge station. In a previous section we have shown how we got the value of 0.4 km for the TU/e. The calculations will be based on a single charging station which will be enough for areas like the TU/e. The "height" option defines how high the drone will fly. In general the drone will be able to see more at once when flying at a higher heights, however as will be discussed later on there are restrictions with respect to law and also on the distance at which people can clearly be distinguished. The "speed" option defines the default speed at which the drone will fly and the "time frame" option will define in what time the drones should be able to cover the entire campus. Based on this time frame the amount of drones needed will be calculated. The "tilt angle" corresponds to the alpha angle from the computations above and defines the angle at which the camera is looking at the ground.

Finally some text boxes are in place that can be used to define some cost values. In the case of the TU/e the cameras cost about 3500 euros based on the interview with the head of security. The next option defines how many cameras will be replaced. On the TU/e campus there are about 50 cameras in total. Next we have options to define the yearly salary of an employee and how many should be replaced. Right now there are always 3 security employees at the TU/e, each with a yearly salary of about 50000 euros. The last option is to define the software cost that comes with programming the drones.

To the right of these settings the output of the calculations can be seen. At the top some values for the particular drone used are shown. These values include drone cost, battery life and field of view of the camera for example. The next value, the coverage, is the percentage of the campus that is seen by the drone within the timeframe. In this case this is 20.99% of the campus. From this percentage we calculate that we need 5 drones to see the entire campus in 15 minutes. The next value shown is the uptime, which is based on battery life and wheather effects. Here the drone not being waterproof has the biggest impact. The next couple of lines display different costs. The main thing to look at is the "Total additional costs for 10 years" line. This line says how much more expensive (negative value means cheaper) it is to deploy the drones. In this case it costs 41065.07 euros more than the current situation.

We can clearly see that currently the salary of the security employees is the biggest part of the cost. If we would be able to replace an employee we would be able to deploy the drones without any issue in this situation. However it may not be feasible to replace an employee since there are only three (at the same time).

Conclusion

Using this tool we can find certain break even points with respect to the cost. When using the DJI phantom 4 pro drone at a height of 25 meters, with a speed of 10 km/h, a camera tilt angle of 30 degrees and a total coverage duration of 15 minutes we can find that we need to replace 34 cameras in order to have no extra costs over a period of 10 years. We consider these settings to be close to the actual settings that will be used eventually, changing the height or the angle can have a big impact on the cost, but flying to high or to low or looking at the ground at a high angle will make the system harder if not impossible to use, as you will not be able to see what is going on. We think this height and angle are balanced, such that we can cover a lot of ground, but don't lose functionality. The main thing to test experimentally is at which speed the video feed of the drones can still be adequately checked by employees, our tool obviously concludes that the higher the speed of the drones the less drones we will need. Replacing about 34 cameras can be a problem when using the phantom 4 since the uptime of this drone is only 40.87%. This would mean that during the other 59.13% of the time we would still have to rely on the cameras. When we replace 34 out of the 50 we will lose too much coverage of the campus.The low uptime of this drone (40.87%) is not that high due to the wheather effects discussed above. The DJI Wind on the other hand has an uptime of 99.33% because of it's great wind and rain resistance, when we want to replace cameras we need this greater uptime. If we were to use the same settings as with the previous calculation and replace all 50 cameras we find that there is an additional cost of 276004.87 euros. If we take it to the extreme, where we fly at 30 meters and have a camera tilt of 45 degrees we see that we still have an additional cost of 125669.91 euros with respect to the current situation (Though in this situation the drones might not be as effective due to the high height and large angle). This big increase in cost can be explained by the bigger uptime, since when we fly all the time we will use more energy which will cost more money. The DJI Wind 1 itself also costs more money, 10 times more than the phantom 4. However this is because the Wind 1 is a custom made drone, in the future when more wheather resitant drone will come on the market this increase rate will go down. In the first case of the DJI Wind 1 we see that both the energy cost and the drone cost amount for about 50% of the total costs. We found a drone, the HexH20 pro V2, that is cheaper in cost. This one only costs 4 times as much as the phantom 4, however it uses way more power. Using the same settings we see that we have an additional cost of 675385.86 euros, which is more than the extra 276004.87 that we got for the Wind 1.

In the calculations above we have not considered yet to replace employees, if we could replace a single employee we would be able to introduce the Wind 1 drones without any issue and still save about 200000 over a period of 10 years. However we should test experimentally to what extend we can replace employees, since this might be difficult as there are only 3 at the moment.

In the calculations above we left the software costs out, since this cost is the hardest to estimate. This cost is usually underestimated and can be rather big. We estimate this cost to be between 10000 - 150000 euros, meaning that we will definetely have to replace at least a single employee if we want to introduce the technology without any extra cost. With the remaining money we can keep at least 10 cameras, meaning that we still have some reliable reference points of the most important areas of the campus.

he found break even points are not really a realistic setting as the TU/e security was not open to replacing employees and replacing almost all cameras is probably not a good idea either, since the largest uptime for the drones we used was 91.29%, so that would mean that the TU/e security would be blind for about 10% of the time, which is unacceptable. This is why we've done a calculation for a situation we think is realistic for each drone type. We used the following settings for this calculation:

  • Height: 25m
  • Speed: 10 km/h
  • Tilt angle: 30 degrees
  • Amount of cameras replaced: 20
  • Amount of employees replaced: 0
  • Software costs: 150000

We chose these settings, such that we don't fly to high and lose visibility, we don't fly to fast and get blurry video footage, we don't have a too large angle so that we can still see what's below the drone and we keep enough cameras to not be blind when the drones can't fly. Using these settings we got the following additional costs:

  • DJI Phantom 4 Pro: 200882.08 euros
  • DJI Wind 1: 542575.75 euros
  • HexH2O Pro V2: 941953.87 euros

While the DJI Phantom 4 Pro is the cheapest option, it can only fly 40.87% of the time, due to not being waterproof. We don't think this would be a viable option for securing the TU/e, So we would rather go for the DJI Wind 1, which is the second cheapest option. Still this seems to cost too much for the TU/e security.

Even if this drone becomes cheaper in the future (it costs 14950 euros now) It would still be too expensive due to the costs of electricity. Lets say it will cost 5000 euros in the future, the additional cost will still be about 300000 euros, which is probably still too expensive.

So to conclude, financially it is not feasible to implement drones for security at the TU/e, if we don't replace any employees. If the TU/e security would agree to replace 1 employee for the drones, it would be feasible, but the TU/e security really didn't like this idea, so we don't think that will ever happen.

Law analysis

This chapter describes the legislation regarding filming humans and flying drones. First the current national legislation is described. Since the European Commision is working on laws for all countries in the European Union, their plans for the future as presented by them in the prototype regulation for unmanned aircraft are also described.

Current legislation

Filming humans

Currently, cameras film the campus everywhere. Cameras have a very good view and can even spot particular scenes inside buildings on campus. The footage is watched by security people in the control room and they decide whether action has to be taken. Footage is preserved for 7 days and then automatically deleted. In case there has been reported a crime, the relevant footage is filtered and used for security objectives. This footage is locked and only shared with responsible and legal authorities.

In case of drones instead of standing cameras, the footage will be treated equally. It will remain available for 7 days and after that all is deleted. Without detected anomalies, nobody sees the footage anymore.

Thus, the subject of privacy is comparable to current standards. Nowadays, all footage is filmed and analyzed and this will be the same situation using drones. Privacy will not be violated more when making use of drones as people can always be spotted now as well.

Flying drones

Legislation about flying drones can be found in "Regeling modelvliegen"[7], which is about aircraft that are not able to carry a human, and are strictly used for demonstration, recreation, or sport.

Where can you fly drones ?

Certain areas are restricted for drones for example near airports. You are only allowed to fly drones in airspace category G, which is the free airspace. Sometimes there is a temporary restriction for example at big events. The red areas in the following image are restricted zones.

Areas.PNG[8]

Apart from flying in airspace category G, there are some more general rules about flying drones:

  • You have to give way to all other aircraft.
  • You have to fly at a safe distance from people and buildings.
  • You always have to be able to see the drone.
  • You are only allowed to fly in daylight.

Rules for private use of drones

[9]

For private use the following rules hold:

  • The maximum height you can fly the drone at is 120 meters.
  • You are not allowed to fly above certain areas such as highways, railways, and ports.
  • Anyone can fly the drone, you do not need a licence.
  • Weight of the drone must not exceed 25 kilograms. (Will be changed to 4 kilograms)


Rules for business use of drones

[10]

The following rules hold if you want to use a drone to generate income. For example video production companies filming using a drone equipped with a camera, or inspection companies that want to inspect places that are difficult to reach.

If you want to use drones for business, you need a permit. There are different kinds of permits, which one you need depends on the situation. If you are the pilot, you need a pilot license. The company that wants to use the drone needs a RPAS Operator Certificate (ROC). There are two kinds of ROC’s, ROC and ROC Light, which one you need depends on the weight of the drone and the height or distance you want to fly the drone at. The owner of the drone needs a proof of airworthiness and a proof of registration to the aviation register.

For business use the following rules hold:

  • Weight of the drone must not exceed 150 kilograms. (ROC)
  • Weight of the drone must not exceed 4 kilograms. (ROC Light)
  • Distance to the pilot must not exceed 500 meters. (ROC)
  • Distance to the pilot must not exceed 100 meters (ROC Light)
  • The maximum height you can fly the drone at is 120 meters, but you can ask for permission to fly higher. (ROC)
  • The maximum height you can fly the drone at is 50 meters. (ROC Light)
  • You are allowed to fly near certain areas such as highways, railways, and ports, but you have to keep a certain distance. You can ask for permission to get closer than the specified distance. (ROC)
  • You are allowed to fly near certain areas such as highways, railways, and ports, but you have to keep a certain distance. (ROC Light)
  • The pilot needs a proof of competence (RPA-L) and has to pass a medical test. (ROC)
  • Anyone can fly the drone, you do not need a licence. (ROC Light)
  • The owner of the drone needs a proof of registration to the aviation register. (ROC)
  • The operator should be at least 18 years old. (ROC and ROC Light)


Penalties for breaking the rules can be: fines, confiscation, or time in jail.

How can privacy be preserved

Footage of the camera is stored only 7 days. Without any report of anomalies, the footage is deleted forever. In case people detect anomalies, the relevant footage is securily distributed to the parties involved (government security, police). Nobody has access to these images but the head of security, all footage is locked safely.

Obviously, safety is considered more important than privacy in case of an offense.

Plans of European Commission

[11]

The European Commission believes unmanned aircraft is a sector that is developing fast, and has a lot of potential for growth and to create jobs. In the next 20 years the drone sector in Europe is expected to employ more than 100,000 people, and have an economic impact of more than 10 billion euros per year. [12] "Unmanned aircraft" includes large aircraft, which resemble in terms of size and complexity manned aircraft, but also small consumer aircraft. Especially those smaller consumer aircraft are being used more frequently in the European Union. However, the rules are different for each country and often not everything is addressed.

Therefore the European Commission is working on new rules regarding drones. They define three categories of unmanned aircraft, each of these categories has different safety requirements which are proportional to the risk of the drones in that category. The categories are:


  • “Open” (low risk): no authorization required to operate the aircraft, as long as forbidden or restricted areas, defined by the National Aviation Authority (NAA) are respected. Safety is ensured by operational limitations, mass limitations, product safety requirements, and a set of operational rules.
  • “Specific” (medium risk): authorization required by a NAA, a risk assessment has to be performed by the operator. For certain lower risk scenarios, a simple declaration sent by the operator to the NAA will be sufficient.
  • “Certified” (higher risk): requirements are similar to those for manned aviation. The process is comparable the process for manned aviation, oversight by NAA and European Aviation Safety Agency (EASA) will be required.


A prototype regulation[13] was presented in the summer of 2016. The document is a prototype regulation for the operation of drones in the open and specific categories. It does not constitute any formal commitment, its purpose is to inform and consult stakeholders. Feedback will be gathered which will be used to develop the actual regulation. It represents the current views of EASA. There is also an explanatory note[14] which provides the rationale and background for the prototype regulation. This regulation is supposed to enter into force in 2019.

Summary of Prototype Regulation

The prototype regulation addresses the “open” and “specific” categories.

Regulation should be proportionate to the nature and risk of the type of drone operation. It should take into account: the type of operation and whether the operation is open to the public; how dangerous the operation is to other air traffic, or persons and property on the ground; the type of airspace used; the complexity and performance of the aircraft; the type, scale, and complexity of the operation that is performed.

The risk of operating a drone varies depending on the characteristics of the drone and the type of operation. Therefore there should be different rules for different categories, based on risk assessment and performance.

The higher risk operations of unmanned aircraft should regulated similarly to manned aircraft, which includes the certification of the aircraft.

The lower risk operations should be regulated by less strict requirements. These operations will be subdivided into the “open” and “specific category”. Risk assessments should be conducted, in order to reduce the administrative burden for operators and authorities. Standard scenarios should be identified.

Both categories

The following rules apply to both categories.

Before the operation the remote pilot shall familiarise themselves with the area the drone will be flying in. This also includes the locations of people, properties and other hazards. Furthermore the pilot shall obtain updated information about flight restrictions or conditions. These are published by the competent authority that may be relevant to the operation. The pilot has to make sure the unmanned aircraft is in a safe condition, and has been updated with geofencing data and recovery procedures. Finally, the operator has to be familiar with the operations manual for the operation, and ensure that the flight conditions are compatible with the authorised conditions.

During the flight the remote pilot shall avoid reckless manoeuvres with the drone and ensure safe operation of the drone. The pilot must comply with the limitations defined by the competent authority, and discontinue a flight when not doing so would be dangerous. Furthermore the pilot must operate the unmanned aircraft within performance limitations defined in the user manual. Finally the pilot must not use the unmanned aircraft to transport goods or passengers, not fly close to emergencies and respect people’s fundamental rights.

Open Category

Operation of unmanned aircraft in this category will fall into one of the following subcategories:

  • A0: Operation that poses a negligible risk of severe injury to people on the ground or damage to manned aircraft. Requires neither specific pilot competence nor age limitations. Operations in this category should be conducted up to a height of 50 meters above ground level. Operators should maintain visual line of sight and the drone should not exceed a horizontal distance of 100 meters from the operator. The drone itself should have a weight of at most 250 grams, and a maximum speed of less than 15 meters per second.
  • A1: Operation complying with requirements ensuring that they pose a negligible risk of severe injury to people on the ground or damage to manned aircraft. Requires neither specific pilot competence nor age limitations. Operations in this category should be conducted up to a height of 50 meters above ground level. Operators should maintain visual line of sight, and be at least 14 years old. The drone should have a weight of at most 25 kilograms.
  • A2: Operation complying with requirements ensuring that they pose a limited risk of severe injury to people on the ground or damage to manned aircraft. Operators should be registered, and equipped with geofencing (virtually defined geographic boundaries) and electronic identification. Operations in this category should be conducted up to a height of 50 meters above ground level. The operator should maintain a minimum horizontal distance of 50 meters from uninvolved persons. Operators should maintain visual line of sight, be at least 14 years old, and have appropriate familiarisation according to the user manual of the drone. The drone must be equipped with geofencing and electronic identification systems. The drone should have a weight of at most 25 kilograms.
  • A3: Operation complying with requirements imposing technical mitigations, like geofencing and electronic identification. These operations pose a higher risk of severe injury to people on the ground or damage to manned aircraft, and are operated by registered operators with higher competence. Operations in this category should be conducted up to a height of 150 meters above ground level. The operator should maintain a minimum horizontal distance of 20 meters from uninvolved persons. Operators should either maintain visual line of sight, or have an observer in visual line of sight. Operators should also be at least 14 years old, carry evidence of their competence of flying unmanned aircraft, and have the appropriate familiarisation to minimise the risk to third parties.. The drone must be equipped with geofencing and electronic identification systems. The drone should have a weight of at most 25 kilograms.

Operators should register in the member state where they have their permanent residence, except when operating in A0 or when they are already registered in the “specific” category.

Specific Category

Operators should submit an operational declaration to the competent authority. The EASA shall provide standard scenarios, for which a risk assessment has been conducted, and mitigating measures have been identified.

If the operation does not correspond to any of the standard scenarios, the operator shall conduct a risk assessment and identify mitigation measures in order to limit the risk to an acceptable level. The operator should consider key factors such as operational area and conditions, and the category of airspace. The operator should also provide an operations manual adapted to the type of operation.

After receiving an application for operational authorisation, the competent authority verifies that the application contains all the required information and documentation. The authority then decides whether or not they give an authorisation. The authorisation can be for limited or unlimited duration. This authorisation can later be suspended, revoked or amended if the operational conditions change.

Conclusion

We do not think filming people will be a problem when deploying drones with cameras. If we treat the footage of the drones equally to the current situation with cameras, the subject of privacy will be comparable to the current situation. Privacy will not be violated more than it is currently, with the use of security cameras.

There are however multiple rules that prevent the use of drones as described in the scenario.

First of all the remote pilot has to have visual line of sight of the drone. We intend to let the drones fly around the campus autonomously. This rule limits the routes the drones can fly, as they cannot fly behind buildings because the operator will lose line of sight. This rule doesn't necessarily prevent the use of drones, but it largely removes the benefit of using them over security cameras.

Secondly, Eindhoven is in a no-fly zone. This is the main rule that prevents the use of drones for security. Currently you are not allowed to fly in Eindhoven because of the nearby airport and military base.

Finally, you are only allowed to fly drones in daylight. This rule also does not necessarily prevent the use of drones. However, if you cannot use the drones approximately half of the time, it is probably not worth it. In order to secure the area the other half of the time security cameras would still be necessary. Then our idea to replace cameras by drones does not work, because you would still need cameras for security during the night.

In order for our idea to work these rules have to be removed, or an exemption has to be made for certain organisations. This seems unlikely though, as the prototype regulation presented by the European Commission also contains these rules.

Societal Analysis

The implementation of drones in daily, security TU/e campus life will be a social matter as well as a technological one. We decided to analyze three visions on safety guarantee by drone implementation, namely the perspective of:

- TU/e Security

- Police Office, Community of Eindhoven

- Students living on the TU/e campus

We chose to interview the Head of Security TU/e and a police officer specialized in drone authorization, community of Eindhoven Noord-Brabant. We asked students to fill in a survey as this would allow us to reach a broad group of people.

Interview

TU/e security

On the thirteenth of March, we met Mr. Becks, who is in charge of the security of the TU/e campus. He explained that the TU/e security is serving the government and thus Eindhoven in general. TU/e is not the party that is in charge on level of its security. Below, we report in short our conversation and the relevant information.

Privacy

- We know that TU/e has a licence to surveil people using cameras. Does that apply for surveilling using a drone carrying the camera?

Yes, the same legislation with respect to privacy can be followed. When entering the campus, everybody can read signs that the terrain is filmed. Besides, the footage will be stored for 7 days and then deleted. In case of reported crimes, the relevant footage is saved and used by authorities. All the footage is safely locked.

Legislation thus depends on the government. Drones must remain within the boundaries of the laws of Eindhoven. Unfortunately, radar can be obstructed by drones and thus planes can have difficulty in landing etc. Important other topics are height of flight, sounds of the drones, and line of eyesight.

Financial

- How many guards are working at the same time (different at different times)?

The TU/e security team consists of 12 people who are surveilling in shifts of 3 persons. The control room is manned 24 hours to check footage for anomalies. In case an anomaly is detected, guards are sent towards the specific position. Besides, 10 guards are hired from external companies to help the TU/e team.

Every evening the team closes the buildings (from 18.00 to 22.00) and every night there are shifts to check the buildings (at 23.00 last of the day). The use of drones will be most efficient outside buildings. Besides, the replacement of people by drones is not supported as drones do not have sufficient artificial intelligence to make rational decisions.

- How many cameras are there?

240, including indoors. Outside, there are almost 50 cameras and most of them are 360 degrees. Inside, cameras are not turning 360 degrees. The footage of indoor cameras is not that good but drones indoor will not be accepted as this will actively violate privacy. Besides, the building structure does not allow flying objects moving freely.

Outside, crimes start so it is most important to surveil the surroundings of the buildings.

- Cost of the camera's

The camera itself, without supporting pole, is around 3500 euros, minimally. This is either a 360 degrees camera or a normal one.

- Approximate coverage of cameras

Most cameras cover the 360 degrees and 100 meters sight. All depends on the weather. In case of a very clear sky, one could see contours that are 15 kilometers away. Cost of maintenance of cameras They have a lifespan of 10 years where no maintenance is really needed. After those 10 years, they will be replaced by new cameras that are improved a lot technically.

- Costs surveillants

Surveillance guards serving TU/e security (fte) have an income on the scale of 50.000 euros. This considers 12 man, 40 hours a week. Besides, around 300.000 euros is yearly spend on hired staff, supporting the TU/e guards.

Law

- Rules regarding flying drones on TU/e campus

The legislation that is binding on TU/e campus, is restricted by the Eindhoven government. These laws are based on laws for gliders. Eindhoven, and thus the campus, is a nofly-zone as the drones can influence the radar of the military base. Actually, when the correct guidelines are created, mr Becks believes that the campus could function as a permitted fly-zone.

Technological

- Is post processing applied to the camera feed (human detection) or is raw feed shown to personnel?

Raw material of the footage is directly shown to surveillance guards, manning the office. After this, footage is saved for 7 days and then deleted completely. The saved footage is locked and Mr. Becks is the only one with a key. Guards who watch the footage have to stick to an ethical code and contract they signed regarding privacy.

The hacking of drones is possible and thus footage can be stolen. On the other hand, cameras have been stolen in the past and hidden cameras are placed on the campus. Security knows that strangers can make images on their campus but they cannot prevent it. Thus, there are always risks.

Innovation and change

- Does the TU/e Security want to use drones?

Yes, they support innovation but the current cameras are good enough. The footage is clear and sharp. Besides, people are rather slow and drones are fast: this can improve surveillance. There live 15000 to 16000 people on campus, so this needs enough protection. Before drones can be implemented, a solid argumentation is needed.

Crimes

- Are all crimes noted and pursued?

No, small crimes are ignored, like theft of TU/e property or so.

- Can you give an indication of damage by theft?

Every year, around 50.000 euros worth of laptops is stolen.


Police Office, Community of Eindhoven

On Monday the 20th of March, we spoke to an officer of police faculty Eindhoven-Noord. He gave the general view of the police in Eindhoven considering the use of drones and the possibilities. It was rather difficult to get in contact with the right people and nobody actually took responsibility. This can be explained by the fact that the police is not that used to innovative technology: drones are no focus point in general. Below, we provide a summary of the relevant details of the conversation.

The use of drones of TU/e campus is currently not possible. Eindhoven is a ‘no-flyzone’, which means that it is forbidden to use drones here. The use of drones in private cases is possible and this could be of use considering TU/e. When you want to use a drone as a company, you need official authorization of the government, city of Eindhoven. The certificate that is needed to get authorization follows a training. This training learns people how to fly drones and assures that the people who will use the drones, are qualified. Nevertheless, this training costs 1500 euros and the use of the drone will still be restricted to a specific location and timespan.

The no-flyzone of Eindhoven is caused by the stationing of a military base nearby. Besides, drones in the air can distract pilots by influencing their radar signal. However, the police are experimenting with the use of drones for surveillance irrespectively of the restricting laws that prevent them from practical use. It is important to be prepared when legislation changes and government will allow use of drones. Situations where drones could come in handy, are cases where large groups of people gather. At the same time, these moments increase danger as failure of the drone will harm a lot of people at the same time.

In conclusion, police is testing the possibilities of drones. Currently, Eindhoven city is restricting the use of drones for surveillance due to risk of technical errors and the negative influence on radar. To use drones for security on the TU/e campus, legislation has to be changed and the drones that are used must be of high quality to limit damage risk. In the future, drones can be implemented but for now their use is restricted as the technology is not yet dependable enough.

Survey

We deployed this survey to the students living on the campus (main goal) and studying in the TU/e (secondary goal).

Analysis of results

In case of the government and police related interview, we see that using drones is not an option under current circumstances. Eindhoven is a restricted area regarding drones due to the military base and aeroport. For incidental use, authorization can be realized but this costs a lot of money and time as a specific drone flight certificate is needed. In the future, when current legislation will be more flexible, the use of drones on campus could be an option, assuming failure risk is neglectably low.

The interview with the head of TU/e security resulted in the fact that current surveillance is very sufficient. Of course, imporvement is always welcome but replacing people by drones would deminish human ratinality observing real life incidents. When using drones instead of guards, decision making is placed into the hands of the people in the surveillance center. This makes immediate action, when detecting anomalies, not even optional any more. Besides, current cameras can watch all TU/e grounds and using drones would not increase the view on the terrain. In short, replacement of current facilities by drones is not necessary. However, adding drones to the current team seems an option as drones are fast and the TU/e supports innovation.

We conducted a survey and asked people living on campus (Aurora and Luna building) and students in general what they think about the idea. In total we received 35 answers out of more than a 1000 people asked. Now we will analyse their responses.

We conducted a survey and asked people living on campus (Aurora and Luna building) and students in general what they think about the idea. In total we received 35 answers out of more than a 1000 people asked. Now we will analyse their responses.

1.The first question we asked the students is: “Do you think that TU/e campus provides a safe environment for students?”. Here are the results:

Screenshot 1.png

Students in general think the environment on the TU/e is very safe, as you see above students could rate how safe they feel on the TU/e from 1 to 4. Where 1 is considered very bad and 4 is very good. Most students picked 4 and feel very safe on campus, the rest all picked 3. We can conclude from this that people feel safe with the current security.


2. The second question we asked all the people who took the survey was: “Have you or someone you know ever had experience with criminals on campus (theft, violence, vandalism, etc.)?” These are the results: 77% answered “No” and 23% said they or their friends experienced a bike theft. We did not get any other criminal activity experiences, so out of this we can conclude that TU/e campus is a fairly safe place for students.


3. The third question we asked the people was: “If yes, was the issue resolved (found stolen belongings/person responsible)?” Out of those 8 cases of bike theft, only 1 case was resolved, so the catch rate was only 12.5%. We were surprised by those results since we expected a higher catch rate.

4. The fourth question we asked was “Do you think security should be improved?”. In the graph below you see the results of the question: Do you think security should be improved? 68.6% of people answered no here, which means most people think security as fine as it is. Still almost 3 out of 10 people have other thoughts about this, 22.9% say Yes here and 8.6% say “Other”, which were in general: It’s good as it is but it always can be improved, and 1 perhaps.

Screenshot 2.png

5. The fifth question we asked was: “How do you think the security be influenced, if we substitute some camera's by surveillance drones?”. Here are the results:

Screenshot 3.png


This is a scale from 1 to 3 where 1 means that security will be hindered and 3 means that it will be improved.

6. The sixth question was: “Would you mind if there would be autonomous drones flying around campus at all times?”. Here are the answers:

Screenshot 4.png

By “other”, people mostly said that they might mind in the beginning or were not sure about it.

7. How much do you like this drone idea?

Screenshot 5.png

In this question we asked how they liked our idea. The students could rate this from 1 (Don’t like it at all) to 6 (I really like this idea). The results were really diverse and it seems people mostly have a strong opinion about this, either they really liked it or they really didn’t. We were surprised to see that so many people didn’t like the idea of security drones.

End results

What we can conclude from this is that people are very satisfied with the security at this moment. When we asked them about deploying drones the results were very diverse. We were surprised that so many students were pessimistic about drone security. There were more students which didn’t like the idea (57%) compared to the students that did like the idea (43%). Maybe in the future, when drones become more familiar to the public and do not disturb them as much, people would accept them easier.

Conclusion

By means of conclusion, looking back at our objective questions, at this point in time it is not financially feasible to replace cameras and add drones for surveillance on campus. We derived those results by utilizing our financial model which we created with respect to all relevant laws and rules concerning drones. We found that filming people on campus is not going to be an issue for TU/e security, since they have a licence for filming anyway. Despite that, flying drones autonomously is an issue because current legislation states that a person should always have the device at a line-of-sight distance. In addition, we conducted interviews with the head of security of the Technical University on one side and a police officer specialized in drone authorisation on the other. Finally, we found out that most people are sceptical of the idea by conducting surveys. Society is not at that point of familiarity with drones to let them fly autonomously and film their every move.

Discussion

Model:

If we would have had more time, we could have considered more reliable data by conducting experiments with actual drones and people and reduce the number of assumpitons we now have. For example, the time needed to change the drone's battery is one piece of data which we estimated by ourselves without having a drone to measure the time with. Moreover, the difference in area surveilled by a drone moving in a straight line compared to moving in other directions (dodging obsticles, change of direction, etc.) is measured by a variable which we do not know if it is accurate right now. The only way of knowing that is by experimenting with actual devices. Another point of improvement is to take into account the ability of operators to detect anomalies using moving cameras (drones) compared to static cameras. Also at what survellance drone speed does the anomaly detection abilities of the person watching the video hinder. Besides, we could take into account the impact of future technology development. Price of technology decreases so drone options will be more realistic. This transitions can be included in the model.

Survey:

We tried to keep the survey as short as possible because we thougth that people would not fill it otherwise, but as a point of improvement we could have included a detailed explanation of the idea and why it is good. The lack of such details might be the reason why a large part of the people who took the survey were not a fan of the concept.

Weekly logbook

Week 1 Starting here, the project was introduced and we formed groups. To generate an innovative and feasible idea, we held several brainstorm sessions. Coming up with a realistic approach to cure a relevant problem using technology, we shifted from several perspectives. In the end, we formulated a topic of research that entailed both security issues and privacy legislation. The basic ideas were covered in a presentation, to be given in week 2. We tried to create slides that were concise and complete, without too much text and with use of visuals.

Week 2 The presentation was completed and we started on the Wiki page. We figured that the regular update of the Wiki will contribute strongly to our progress. Besides, this logbook gives a good overview of our objectives and the aspects of the project that were already covered.

On Thursday, the presentation of the chosen project challenge will be given. Feedback from the other students and the expert panel can be implemented.

Week 3 After getting feedback on our first presentation, we had to change important aspects of our research plan. The most important parts that needed attention were stakeholders, privacy and benefit. We devided tasks, related to personal skill. You either work on programming or on the social/public part that considers law. In the future, it will be important to keep close communication between group members to ensure that everybody understands the progress in the two directions. It is obvious that the technology and society must work together to realize success.

Besides deviding tasks and narrowing down the procedure, time management was a very important part of this second week. The planning of the project needed a lot of time but it is a precondition for success. Now we know what, when and how to deliver parts and when we stick to the Gantt chart, we will realize an optimal research project.

We presented our time planning including the deadlines and tasks we defined. This is an important moment for feedback, as this is the last moment to check your project planning and idea in public. Something else that needed attention, was our resource allocation. It was important to check whether our ideas were feasible. After all, we planned out a detailed project and included our feedback.

Week 4 The division of tasks, created together with the Gantt chart, was executed this week for the first time. On Monday, the group was split in two, where three of us started to create the financial model and try the drone while the others focused on law and legislation. This second group also tried to contact the police office. We suspected that officers would have a very realistic view on our project and their opinion should be taken seriously. Unfortunately, one of our group members had an accident so we were working with five for the majority of this week.

On Thursday, we planned a feedback session with our expert panel. Learning goals will be discussed and evaluated.

Feedback on Thursday was mainly focused on following parts: - The target groups for our surveys; - Creating the boundaries for our research, mostly defining an area (e.g. the TU/e campus); - Formulating scenarios to present to the participants in questionnaires; - Find out what the security legislation is with regard to the chosen area; - Check European Commission legislation

Based on these instructions, we chose TU/e campus as target area and we divided tasks again. Two of us (as someone was injured) focused on the financial model to find out what the current costs are when using a drone. Two of us focused on the European Commission and it's legislation concerning privacy and the use of drones. One of us contacted the head of the security department of the TU/e to make an appointment. Besides, a document was created to inventarize all the relevant questions for the head of security. As the current situation is the starting point for development, these questions deserve a lot of attention.

Week 5 This week, we managed to create a solid base for our future studies. We arranged a meeting with the head of the TU/e Security section, mr Becks. He could answer all of our questions concerning surveillance and security on campus. From this point, we had enough information to develop a more reliable and realistic plan.

Also this week we started on the requirements document, we stated what our drone should be able to do and also stated some safety requirements.

On the financial model part, we defined more boundaries concerning the model:

Besides, we concluded that the police office is not that easily reached.


Week 6 This week, we continued working on the financial model, we determined what variables should be in the model and we made functions for some variables (for example the height vs line of sight. We also started to implement this in java so we can set some unknown variables.

Also we started looking for a drone that fits our requirements, as for now it seems that the Phantom 4 Pro fits our requirements the best.

For the law part we are looking into the European Prototype regulation which means the laws for drones might be changed in Europe, we don't only want to look into the current law but also what it could become and what are the possibilities then.

Week 7 During this week we made the survey and asked hundress of students (most living on campus) about their opinion. We got 36 replies and analysed the results.

Furthermore, we finalized the business model and worked on some bug fixes.

Also we did work on this wiki, adding and changing different sections (for example: made our objectives clearer and made sure we answered all the needed questions).

Week 8 Final presentation and demonstration

References

  1. [1] Area calculator
  2. [2] Area of a trapezoid
  3. [3] Average distance to center of circle.
  4. [4] DJI phantom 4 specs
  5. [5] KNMI wind statistics
  6. [6] Amount of dry days in the Netherlands
  7. [7] Regeling modelvliegen, Dutch government
  8. [8] Drone no-fly zones
  9. [9] Rules for private use of drones, Dutch government
  10. [10] Rules for business use of drones, Dutch government
  11. [11] Civil Drones, EASA
  12. [12] Unmanned aircrafts, European Commission
  13. [13] Prototype regulation drones, European Commision
  14. [14] Explanatory note for prototype regulation drones, European Commision