No description available.

Landing a rotorcraft unmanned aerial vehicle (RUAV) without human supervision is a capability that would significantly broaden the usefulness of UAVs. The benefits are even greater if the functionality is expanded to involve landing sites with unknown terrain and a lack of GPS or other positioning aids. Examples of these types of non-cooperative environments could range from remote mountainous regions to an urban building rooftop or a cluttered parking lot. The research of this thesis builds upon an approach that was initiated at NASA Ames Research Center to advance technology in the landing phase of RUAV operations. The approach consists of applying JPL's binocular stereo ranging algorithm to identify a landing site free of hazardous terrain. JPL's monocular feature tracking algorithm is then applied to keep track of the chosen landing point in subsequent camera images. Finally, a position-estimation routine makes use of the tracking output to estimate the rotorcraft's position relative to the landing point. These position estimates make it possible to guide the rotorcraft toward, and land at, the safe landing site. This methodology is implemented in simulation within the context of a fully-autonomous RUAV mission. Performance metrics are defined and tests are carried out in simulation to independently evaluate the performance of each algorithm. The stereo ranging algorithm is shown to successfully identify a safe landing point on average 70%-90% of the time in a cluttered parking lot scenario. The tracking algorithm is demonstrated to be robust under extreme operating conditions, and lead to a position-estimation error of less than 1 meter during a 2-minute hover at 12 meters above the ground. Preliminary tests with actual flight hardware are done to confirm the validity of these results, and to prepare for demonstrations and testing in flight.

A Rotorcraft UAV provides an ideal experimental platform for vision-based navigation. This thesis describes the flight tests of the US Army PALACE pro ject, which implements Moravec's pseudo-normalized correlation tracking algorithm. The tracker uses the movement of the landing site in the camera, a laser range, and the aircraft attitude from an IMU to estimate the relative motion of the UAV. The position estimate functions as a GPS equivalent to enable the rotorcraft to maneuver without the aid of GPS. Flight tests were performed with obstacles and over concrete, asphalt, and grass in daylight conditions with a safe landing area determined by a separate method. The tracking algorithm and position estimation performance are compared to GPS. Accurate time synchronization of the inputs to the position estimation algorithm directly affect the closed-loop stability of the system, proportional with altitude. By identifying the frequency response of each input and adding filters to delay some of the inputs, the closed-loop system maintains stable flight above 18 m above ground, where the system was unstable without the additional filters.

Miniature air vehicles (MAVs) have attracted a large amount of interest recently both from the research community and from the public. New battery technologies as well as rapid developments in embedded processing and MEMS sensor technologies have greatly increased the potential of these vehicles. MAVs have been envisioned playing significant roles in both civil and military applications. Examples include: fire monitoring, search and rescue, traffic monitoring, crop monitoring, convoy protection, border surveillance, troop support, law enforcement, natural disaster relief, and aerial photography. The application of MAVs tends to center on the ability of the MAV to collect and deliver visual information to the user. In many applications it is important to be able to accurately geolocate items of interest in the visual data. However, the inaccuracies associated with MAV platforms have led to relatively large errors in previous attempts at geolocation. The first half of this thesis focuses on increasing the accuracy of geolocation estimates achievable using a hand-launchable MAV. To accomplish this, methods are presented for bias estimation, wind estimation, recursive least squares filtering, and optimal flight path generation. Hardware results are presented which demonstrate the ability to consistently localize targets to within 5 m regardless of wind conditions. The second half of this thesis focuses on using the high accuracy geolocation estimates to complete a search and engage mission. This is a mission in which the MAV not only locates the target, but also accurately delivers a payload to the target site. The focus is on delivering an attached payload via accurate landing at the target site. A vision-based landing approach is presented which is robust to both wind and moving targets. Simulation results are presented which demonstrate the effectiveness of the control.

In Wilderness Search and Rescue (WiSAR), the incident commander (IC) creates a probability distribution map of the likely location of the missing person. This map is important because it guides the IC in allocating search resources and coordinating efforts, but it often depends almost exclusively on prior experience and subjective judgment. We propose a Bayesian model that utilizes publicly available terrain features data to help model lost-person behaviors. This approach enables domain experts to encode uncertainty in their prior estimations and also make it possible to incorporate human-behavior data collected in the form of posterior distributions, which are used to build a first-order Markov transition matrix for generating a temporal, posterior predictive probability distribution map. The map can work as a base to be augmented by search and rescue workers to incorporate additional information. Using a Bayes Chi-squared test for goodness-of-fit, we show that the model fits a synthetic dataset well. This model also serves as a foundation of a larger framework that allows for easy expansion to incorporate additional factors such as season and weather conditions that affect the lost-person's behaviors. Once a probability distribution map is in place, areas with higher probabilities are searched first in order to find the missing person in the shortest expected time. When using a Unmanned Aerial Vehicle (UAV) to support search, the onboard video camera should cover as much of the important areas as possible within a set time. We explore several algorithms (with and without set destination) and describe some novel techniques in solving this path-planning problem and compare their performances against typical WiSAR scenarios. This problem is NP-hard, but our algorithms yield high quality solutions that approximate the optimal solution, making efficient use of the limited UAV flying time. The capability of planning a path with a set destination also enables the UAV operator to plan a path strategically while letting the UAV plan the path locally.

Search operations can be described by the interaction between three entities: the target, the sensor, and the environment. Past treatments of the search problem have focused primarily on the interaction between the sensor and the target. The effects that the environment has on the target and sensor have been greatly simplified or ignored completely. The wilderness search and rescue scenario is one case in which these interactions cannot be safely ignored. Using the wilderness search and rescue problem as our motivating example, we develop an algorithm for planning search paths for a small unmanned aerial vehicle (UAV) over rough terrain environments that provide complete coverage of the specified terrain region while minimizing effort wasted on duplicate coverage. The major components of this algorithm include 1) breaking the search region into smaller sub-regions that are easier to deal with, and 2) planning the search for each of these sub-regions. The original contributions of this thesis focus on the latter of these two components. We use a method based on the directional offset of terrain contours to produce paths on the terrain for the sensor to observe as the UAV follows the flight path. We then employ directional-offset methods again by moving in the direction along the terrain normal from the sensor path to generate a flight path that lies in the air a specified distance away from the points on the terrain that are to be observed. These two paths are linked in a way that provides the sensor with an ample viewing opportunity of the terrain regions below. We implement this planning algorithm in software with Matlab, and provide a complete simulation of a UAV that follows the planned search pattern. Our planning algorithm produced search paths that were 94 to 100 percent complete in test scenarios for several rough-terrain regions. Missed regions for these plans were near the search boundaries, and coverage could easily be provided by subsequent plans. We recommend the study of region segmentation, with careful consideration of planning algorithms as the major focus of future work.

In GIS-projects the data capture is one of the most time consuming processes. Both how to collect the data and the quality of the collected data is of high importance. Common methods for data capture are GPS, LiDAR, Total Station and Aerial Photogrammetry. Unmanned Aerial Vehicles, UAVs, have become more common in recent years and the number of applications continues to increase. As the technique develops there are more ways that UAV technique can be used for collection of geographic data. One of these techniques is the UAV photogrammetry that entails using an UAV equipped with a camera combined with photogrammetric software in order to create three dimensional models and orthophotos of the ground surface. This thesis contains a comparison between different geographic data capture methods such as terrestrial and aerial methods as well as UAV photogrammetry. The aim is to investigate how UAVs are used to collect geographic data today as well how the techniques involving UAVs can replace or be used as a complement to traditional methods. This study is based on a literature study and interviews. The literature study aims to give a deeper insight in where and how UAVs are used today for geographic data capturing with focus on three main areas: environmental monitoring, urban environment and infrastructure, and natural resources. Regarding the interviews companies and other participants using UAVs for geographic data collection in Sweden have been interviewed to get an accurate overview of the current status regarding the use of UAVs in Sweden. Advantages, disadvantages, limitations, economical aspects, accuracy and possible future use or development are considered as well as different areas of applications. The study is done in collaboration with the geographic IT company Digpro Solutions AB. The goal is to be able to present suggestions of how UAV data can be applied in Digpros applications. Information from the literature study and the interviews show that using a UAV makes it possible to cover a large range between terrestrial and aerial methods, and that it can replace or complement other methods for surveying and data collection. The use gives the possibility to get close to the object without being settle to the ground, as well as work environment profits since dangerous, difficult areas can be accessed from distance. The data can be collected faster, quicker, cheaper and more frequent. Time savings occurs in the measurement stage but compared to terrestrial methods more time is required for the post-processing of the data. The use in Sweden is limited due to difficulties linked to Swedish legislation regarding camera surveillance, as well as long waiting times for the permissions that is required to fly. However, a change in the camera surveillance law is expected which means that UAVs will be excluded from the law. That may result in great benefits for everyone within the industry as well as a continued development of the technique and the use of UAVs. / Inom GIS ar datainsamling en av de mest tidskrävande processerna. Både hur data samlas in samt kvaliteten ar av hög vikt. Några av de vanligaste metoderna för datainsamling idag är GPS, LiDAR, totalstation och fotogrammetri. Obemannade flygfarkoster, UAVs, har de senaste åren blivit allt vanligare och användningsområdena fortsätter att öka. I takt med att tekniken hela tiden utvecklas finns idag flertalet satt att med hjälp av UAVs samla in geografisk data. Med kamerautrustade obemannade flygfarkoster och fotogrammetriska programvaror ar det bland annat möjligt att skapa tredimensionella modeller samt ortofoton av markytan. Detta kandidatexamensarbete innehaller en jämförelse mellan terrestra- samt flygburna metoder för datainsamling och obemannade flygburna metoder. Syftet är att undersöka hur UAVs kan anvandas för att samla in geografisk data samt möjligheten att ersätta eller komplettera existerande metoder, samt att presentera en overgripande bild av UAVs anvandningsomåden. Denna studie bygger pa en litteraturstudie samt intervjuer. Litteraturstudien syftar till en djupare inblick i anvandningsområden för UAV tekniken med fokus på tre huvudområden: miljöövervakning, urbana miljöer och infrastruktur samt naturliga resurser. Under intervjuerna intervjuades företag och andra aktörer inom branschen med syftet att göra en nulägesanalys av hur UAVs används för insamling av geografisk data i Sverige. Det insamlade materialet analyserades med avseende pa användningsområden, för- och nackdelar, hinder, kostnader, noggrannhet samt möjlig framtida användning och utveckling av tekniken. Studien är gjord i samarbete med företaget Digpro Solutions AB som är verksamma inom geografisk IT. Målet är att efter studien kunna ge förslag på hur data insamlad med UAV kan appliceras på Digpros applikationer. Information fran intervjuerna och litteraturen har visat att UAV täcker ett stort spann mellan terrestra- och flygburna metoder, och att den kan ersätta eller utgöra ett komplement till många mät- och datainsamlingsmetoder. Användningen av UAVs innebär möjlighet till att samla in data på ett nära avstånd till objekt utan att vara bunden till marken. Den medför även arbetsmiljövinster då farliga, svårtillgängliga områden kan nås från avstånd. Data kan samlas in snabbare, enklare, billigare och mer frekvent. Tisdbesparingar sker i inmätningsskedet men jämfört med terrestra mätmetoder krävs dock mer tid för efterbearbetning av mätdatat. Användningen i Sverige begränsas av svårigheter kopplade till Svensk lagstiftning gällande kameraövervakning, samt långa väntetider på de tillstånd som kravs för att få flyga. Dock väntas en ändring i kameraövervakningslagen som innebär att drönare inte innefattas i lagen. Detta kan komma att medföra stora fördelar för samtliga inom branschen samt en fortsatt utveckling av tekniken samt användningen av UAVs.

This thesis project investigates a method for performing dynamic path planning in three dimensions, targeting the application of autonomous unmanned aerial vehicles (UAVs).  Three different path planning algorithms are evaluated, based on the framework of rapidly-exploring random trees (RRTs): the original RRT, RRT*, and a proposed variant called RRT-u, which differs from the two other algorithms by considering dynamic constraints and using piecewise constant accelerations for edges in the planning tree. The path planning is furthermore applied for unexplored environments. In order to select a path when there are unexplored parts between the vehicle and the goal, it is proposed to test paths to the goal location from every vertex in the planning graph to get a preliminary estimate of the total cost for each partial path in the planning tree. The path with the lowest cost given the available information can thus be selected, even though it partly goes through unknown space. For cases when no preliminary paths can be obtained due to obstacles, dynamic resizing of the sampling region is implemented increasing the region from which new nodes are sampled. This method using each of the three different algorithms variants, RRT, RRT*, and RRT-u, is tested for performance and the three variants are compared against each other using several test cases in a realistic simulation environment.  Keywords / Detta examensarbete undersöker metoder för att utföra dynamisk ruttplanering i tre dimensioner, med applicering på obemannade luftfarkoster. Tre olika ruttplaneringsalgoritmer testas, vilka är baserade på snabbt-utforskande slumpmässiga träd (RRT): den ursprungliga RRT, RRT*, och en föreslagen variant, RRT-u, vilken skiljer sig från dom två första algoritmerna genom att ta hänsyn till dynamiska begränsningar och använda konstanta accelerationer över delar av rutten. Ruttplaneraren används också i okända miljöer. För att välja en rutt när det finns outforskade delar mellan farkosten och målet föreslås det att testa rutten till målpunkten från varje nod i som ingår i planeringsträdet för att erhålla en preliminär total kostnad till målpunkten. Rutten med lägsta kostanden kan då väljas, givet tillgänglig information, även om den delvis går genom outforskade delar. För tillfällen när inga preliminära rutter kan erhållas på grund av hinder har dynamisk storleksjustering av samplingsområdet implementerats för att öka området från vilket nya noder samplas. Den här metoden har testats med dom tre olika algoritmvarianterna, RRT, RRT*, och RRT-u, och dom tre varianterna jämförs med avseende på prestanda i ett flertal testfall i en realistisk simuleringsmiljö.

Unmanned Aerial Vehicles (UAV) are agile and are able to fly in and out of areas that are either dangerous for humans or have complex terrains making ground robots unsuitable. For their autonomous operation, the ability to explore unmapped areas is imperative. This has applications in data gathering tasks, search and rescue etc.  The objective of this thesis is to ascertain that it is, in fact, possible and feasible to use UAVs equipped with 2D laser scanners to perform autonomous exploration tasks in indoor environments. The system is evaluated by testing it in different simulated and real environments. The results presented show that the system is capable of completely and safely exploring unmapped and/or unexplored regions. / Obemannade flygfarkoster (UAV) är smidiga och kan flyga in och ut ur områden som är farliga för människor eller är svårtillgängliga för markrobotar. För att nå höga nivåer av autonomitet måste en UAV kunna utforska och kartlägga ett okänt område på egen hand. Det finns flera tillämpningar för detta, så som räddningsuppdrag och datainsamling. Målet med denna avhandling är att visa attdet är möjligt att använda en UAV utrustad med 2D-laserskannrar för att utföra autonoma kartläggningsuppdrag i inomhusmiljöer. Systemet utvärderas genom att testa det i olika simulerade och verkliga miljöer. De presenterade resultaten visar att systemet kan utforska okända områden på ett säkert sätt.

View planning is an important part of achieving full robotic autonomy. The ability to incorporate the view of a robot as well as the ability of evaluating and choosing the best view are highly desired abilities for a wide variety of robotics applications. In this work we present a new iterative optimization scheme and evaluational method in order to choose the Next Best View in the framework of object modeling. As unmanned aerial vehicles (UAVs) become more and more common we take advantage of the additional degrees of freedom a UAV offers. We show, in simulation, that the proposed method is able to pick out views that are highly relevant in order to model the observed object in question. It is shown that with iterative optimization the resulting view is improved. Additional experiments where the method is deployed onto a real UAV show the real-world applicability of the method. / Planering av hur vyer för bildinhämtning skall väljas är en viktig del för att uppnå full robotautonomi. Att kunna infoga information från olika vyer samt att välja den bästa vyn är viktigt för många olika robottillämpningar. I det här arbetet presenterar vi en ny iterativ optimeringsbaserad metod för att välja bästa nästa vy inom ramen för objektmodellering. I takt med att flygande robotar, s.k. UAVer, blir allt vanligare kan vi utnyttja de ytterligare frihetsgraderna som ett UAV-system ger. Vi visar i simulering att den föreslagna metoden kan välja ut vyer som är relevanta för att modellera ett visst objekt och att vyerna förbättras genom vår iterativa optimering. Experiment på en verklig UAV visar att metoden är tillämpbar i praktiken.