Now-a-days safety systems and their advanced features have become a major part of human lives. People are ready to pay accordingly for the features they get for and very enthusiastic towards technology and latest trends. One such thing is drone or multicopter. These days everybody is getting interested in drones to buy, not only the fact that it is used in various scientific ways, sports and recreation purposes but also the latest advancements that was taking place in the development of light weight flying vehicles has made many scientific researchers, multinational companies and almost all the people to turn their eye towards the development of drones. And many companies are doing research for development of new safety features which can be called as the safety for the future. Some companies already introduced drones into the market and are used in different ways for different purposes. The usage of this vehicles depends on how intelligently one uses these multicopters. This thesis introduces a feature that adds safety to the multicopters to prevent them from flying to no-fly-regions. The work in this thesis is done to provide an approach by the usage of Raspberry Pi 2 B for multicopter applications as the main development board. It also helps the multicopter to prevent entering the NFR by detecting the NFRs around them intelligently and avoid them so there shouldn\'t be any problem or damage for the multicopters. Here we use GPS sensor for getting the NMEA data as input to know the latitude and longitude positions and then transferred to RPI2 B which allows us to know the latitude and longitude positions and then transfer this data into database to store the data through a wireless medium i.e., Wi-Fi medium. Based on the information stored in database we can see the location in a graphical manner using the open street maps (OSM). After that different checks are performed to avoid the NFR : (i) We will check if the current point lies inside or outside the no-fly-region based on the map information of NFR using the Point in Polygon algorithm and then (ii) we are using some area based detection 4 algorithm to check the distance from the point to line using Paul Brouke algorithm to see how far is the next NFR from the current point and avoiding it and the information is updated and stored in the database accordingly .(iii) Later, if the multicopter is out of all no-fly-region then the distance to the next NFR or nearest ones is analyzed and the information will be used for safety purpose. By using geometry and algorithms we are checking and finding out the NFRs and avoid entering into the NFR space. If the point is detected inside a no-fly-region then the last point outside this region will be detected which is marked as safe and the multicopter will be backtracked to the previous point before entering the no-fly-region i.e., the safe point. This paper not only aims at multicopter safety but also throws light into the future systems that are going to be developed in the field of Car-2-X, ensuring extended safety of the passengers.

Now-a-days safety systems and their advanced features have become a major part of human lives. People are ready to pay accordingly for the features they get for and very enthusiastic towards technology and latest trends. One such thing is drone or multicopter. These days everybody is getting interested in drones to buy, not only the fact that it is used in various scientific ways, sports and recreation purposes but also the latest advancements that was taking place in the development of light weight flying vehicles has made many scientific researchers, multinational companies and almost all the people to turn their eye towards the development of drones. And many companies are doing research for development of new safety features which can be called as the safety for the future. Some companies already introduced drones into the market and are used in different ways for different purposes. The usage of this vehicles depends on how intelligently one uses these multicopters. This thesis introduces a feature that adds safety to the multicopters to prevent them from flying to no-fly-regions. The work in this thesis is done to provide an approach by the usage of Raspberry Pi 2 B for multicopter applications as the main development board. It also helps the multicopter to prevent entering the NFR by detecting the NFRs around them intelligently and avoid them so there shouldn\'t be any problem or damage for the multicopters. Here we use GPS sensor for getting the NMEA data as input to know the latitude and longitude positions and then transferred to RPI2 B which allows us to know the latitude and longitude positions and then transfer this data into database to store the data through a wireless medium i.e., Wi-Fi medium. Based on the information stored in database we can see the location in a graphical manner using the open street maps (OSM). After that different checks are performed to avoid the NFR : (i) We will check if the current point lies inside or outside the no-fly-region based on the map information of NFR using the Point in Polygon algorithm and then (ii) we are using some area based detection 4 algorithm to check the distance from the point to line using Paul Brouke algorithm to see how far is the next NFR from the current point and avoiding it and the information is updated and stored in the database accordingly .(iii) Later, if the multicopter is out of all no-fly-region then the distance to the next NFR or nearest ones is analyzed and the information will be used for safety purpose. By using geometry and algorithms we are checking and finding out the NFRs and avoid entering into the NFR space. If the point is detected inside a no-fly-region then the last point outside this region will be detected which is marked as safe and the multicopter will be backtracked to the previous point before entering the no-fly-region i.e., the safe point. This paper not only aims at multicopter safety but also throws light into the future systems that are going to be developed in the field of Car-2-X, ensuring extended safety of the passengers.

Optimal control for multicopters is difficult in part due to the low processing power available, and the instability inherent to multicopters. Deep imitation learning is a method for approximating an expert control policy with a neural network, and has the potential of improving control for multicopters. We investigate the performance and reliability of deep imitation learning with trajectory optimization as the expert policy by first defining a dynamics model for multicopters and applying a trajectory optimization algorithm to it. Our investigation shows that network architecture plays an important role in the characteristics of both the learning process and the resulting control policy, and that in particular trajectory optimization can be leveraged to improve convergence times for imitation learning. Finally, we identify some limitations and future areas of study and development for the technology. / Optimal kontroll för multikoptrar är ett svårt problem delvis på grund av den vanligtvis låga processorkraft som styrdatorn har, samt att multikoptrar är synnerligen instabila system. Djup imitationsinlärning är en metod där en beräkningstung expert approximeras med ett neuralt nätverk, och gör det därigenom möjligt att köra dessa tunga experter som realtidskontroll för multikoptrar. I detta arbete undersöks prestandan och pålitligheten hos djup imitationsinlärning med banoptimering som expert genom att först definiera en dynamisk modell för multikoptrar, sedan applicera en välkänd banoptimeringsmetod på denna modell, och till sist approximera denna expert med imitationsinlärning. Vår undersökning visar att nätverksarkitekturen spelar en avgörande roll för karakteristiken hos både inlärningsprocessens konvergenstid, såväl som den resulterande kontrollpolicyn, och att särskilt banoptimering kan nyttjas för att förbättra konvergenstiden hos imitationsinlärningen. Till sist påpekar vi några begränsningar hos metoden och identifierar särskilt intressanta områden för framtida studier.

This is a master thesis by Maria Ader and David Axelsson, students at the Master of Science in Engineering degree program in Design and Product Realization at KTH, within the master program Integrated Product Design. The thesis work will benefit ÅF and the EU project ɪɴᴛᴇʀᴀᴄᴛ. The ɪɴᴛᴇʀᴀᴄᴛ project is part of the EU’s effort to forward climate research, and aims to “coordinate and harmonize research and monitoring efforts that will greatly contribute to our knowledge and understanding of changes occurring in the arctic environment.” One out of 12 subprojects within ɪɴᴛᴇʀᴀᴄᴛ aims to “increase awareness of drone technology and sensors among researchers and research station managers while making industry aware of innovative potential uses requiring drone and sensor development.” A drone is an unmanned aerial system/vehicle (UAS/UAV), i.e. an airborne vehicle without a human pilot aboard. This master thesis examines the need of drones at the ɪɴᴛᴇʀᴀᴄᴛ research stations and how arctic climates affect drone technology and the ergonomics of piloting a drone. The thesis also provides an overview of the current state of the drone market and the laws and regulations that affect the use of drones. A survey was distributed within ɪɴᴛᴇʀᴀᴄᴛ to map the researchers’ need of, and attitudes towards, drones, followed by exhaustive interviews with researchers and other key figures. Field testing at Tarfala Research Station provided complementing data. The primary insight from the study was that the researchers’ need, as well as the tasks and methods that they employ, vary greatly. Another insight was that many researchers want to use drones primarily as a sensor platform to collect data from large areas in a short time span. A situation-based drone recommendation and a concept proposal for a simple water sampling solution were made based on the results of the study / Detta är ett examensarbete utfört av Maria Ader och David Axelsson, studenter på civilingenjörsprogrammet Design och Produktframtagning på KTH, med masterinriktning Teknisk Design. Arbetet är utfört åt ÅF i syfte att bidra till EU-projektet ɪɴᴛᴇʀᴀᴄᴛ. Iɴᴛᴇʀᴀᴄᴛ är EU:s satsning på klimatforskning i Arktis och syftar till att “koordinera och harmonisera forskning och miljöbevakning som bidrar till vår kunskap och förståelse av förändringar som sker i de arktiska miljöerna.” Ett av tolv delprojekt inom ɪɴᴛᴇʀᴀᴄᴛ-projektet syftar till att öka medvetenheten om drönarteknologi och sensorer bland forskare och föreståndare på forskningsstationerna inom ɪɴᴛᴇʀᴀᴄᴛ, samt att göra drönarindustrin medveten om nya potentiella användningsområden. En drönare är ett obemannat luftfartyg, d.v.s. en flygfarkost utan pilot ombord. Drönare benämns ibland som “UAS” och “UAV”. I den här rapporten används främst den engelska termen “drones”. Detta examensarbete undersöker behovet av drönare på de forskningsstationer som är delaktiga i ɪɴᴛᴇʀᴀᴄᴛ och hur det arktiska klimatet påverkar drönartekniken och ergonomin. Arbetet kartlägger även drönarmarknaden och de lagar och regler som påverkar användandet av drönare. En utförlig studie genomfördes, där forskarnas behov av drönare undersöktes. En enkät skickades ut inom ɪɴᴛᴇʀᴀᴄᴛ och utförliga intervjuer genomfördes med forskare och andra nyckelpersoner. Ett studiebesök på Tarfala forskningsstation kompletterade med fältdata. Den främsta insikten från studien var att behov, arbetsuppgifter och metoder varierar mycket mellan de olika forskarna. En annan insikt var att många ville använda drönare som sensorbärare, och på så sätt insamla data från stora områden på kort tid. Resultatet från studien låg till grund för en situationsbaserad drönarrekommendation samt ett konceptförslag för en enkel vattenprovtagningslösning.

Die vorliegende Arbeit beschäftigt sich mit der Entwicklung eines UAV-basierten Systems und der zugehörigen Methodenentwicklung zur automatisierten Rehkitzsuche in Feldern. Jedes Jahr sterben sehr viele Wildtiere -- vor allem Rehkitze -- während dem Mähen von landwirtschaftlich genutzten Wiesen. Mit herkömmlichen Methoden ist es unter vertretbarem Aufwand bisher nicht gelungen, die Zahl der Mähopfer auf ein erträgliches Maß zu reduzieren. Mit der Entwicklung des in dieser Arbeit beschriebenen "Fliegenden Wildretters" könnte sich das in Zukunft ändern. Mit Hilfe einer Wärmebildkamera aus der Vogelperspektive lässt sich ein warmes Tier, wie ein Rehkitz, wesentlich leichter aufspüren, als mit herkömmlichen Methoden. Auslegung und Aufbau des Systems orientieren sich speziell an dem Aspekt, wie eine möglichst hohe Flächenleistung erreicht werden kann, ohne dabei Tiere zu übersehen. Drei Faktoren sind besonders wichtig, um dieses Ziel zu erreichen: Eine hohe Geschwindigkeit des gesamten Suchprozesses, eine zuverlässige Detektion und eine präzise Lokalisierung der Tiere. Durch Automatisierung lassen sich viele Teilaspekte dieser Aufgabe beschleunigen. Deshalb werden im Rahmen dieser Arbeit verschiedene Methoden entwickelt und validiert, unter anderem zur Flugplanung, Flugsteuerung, Bilddaten-Auswertung, Objekt-Detektion und Georeferenzierung. Die Kenntnis der Rehkitz-Merkmale und der Einflussgrößen bei der Thermalbilderfassung helfen, die Qualität der Detektion zu erhöhen, weshalb sie in dieser Arbeit besondere Berücksichtigung finden. Auch die Präzision der Lokalisierung lässt sich durch Kenntnis der Einflussgrößen auf die Positions- und Lagemessung des UAVs erhöhen. Anhand von umfangreichen Messkampagnen wird die Funktion und Qualität des Systems unter realen Bedingungen belegt.