deployed to the location of an enemy combatant and circles the area waiting for positive identification of the target and clearance. Drones are not allowed to strike if there are women and children nearby, making strikes on residential dwellings forbidden. The drones return to base once a positive strike is confirmed. ............................................................................................................................................. 28 Figure 24 - Interactive map revealing where UAVs are being flown inside the USA . ........................ 29 Figure 25 - A timeline of the rise of the use of UAVs. As can be seen above, UAVs started to be used in serious defense applications in World War II. Their use increased as time went on and after Israel used them to destroy the Syrian air defenses in 1982 the United States decided to start investing heavily in the technology. During the war in Afghanistan the United States had 3,000 drones in service. That number has increased to 12,000 as of 2010 . ................................................................................... 30 Figure 26 - Autonomous Platform for Precision Agriculture (APPA) is an UAV developed by University of Arkansas System. The UAV was developed with purpose of better crop management and nurseries . .............................................................................................................................................................. 31 Figure 27 - An unmannedaerial vehicle developed by NASA in wildfire detection testing . .............. 31 Figure 28 - A UAV launched into a hurricane in 2008. The purpose was to know more about hurricane’s characteristics . ............................................................................................................................... 32 Figure 29 - Launching of RQ-11 Mini-UAV, an UAV that is used for identify possible threats. The UAV acts as second security layer
Both types of communication protocols allow for a wide variety of potential peripherals that can be added to the UAV for custom applications. On the software side, the Matrice 100 allows for customizable flight planning using the DJI Ground Station (GS) Pro application for iOS devices 2 . The iOS device we use to access the GS Pro app is an iPad. The GS Pro app can be used to with the standard DJI remote controller to operate the Matrice 100 at the UAV-standard 2.4 GHz frequency and is capable of 5 km distance communication. The app also provides video downlink from the Matrice 100 camera in order to give first person perspective view from the UAV displayed on the iPad. Most importantly, the GS Pro app provides flight mission planning capabilities which can configure a predetermined flight path with customizable parameters such as UAV speed and altitude. Figure 4.3 shows the standard interface used to plan the mission and adjust the flight parameters. This easy to use interface allows for simple automation of UAV flight without the need for more complex programming systems, such as ROS.
Active (reconﬁgurable) fault-tolerant control system is capable of detecting faults in the system (which are more difﬁcult to detect than failures) and is able to adequately com- pensate for failures (which is more difﬁcult than to only accommodate faults). An active fault-tolerant controller usually contains a separate module: an FDD (Fault Detection and Diagnosis) system that monitors the health of the system. The FDD system provides infor- mation to a supervisory module about the magnitude of the fault, failure or damage. Once the fault is detected, located and identiﬁed (the type and magnitude of the fault), appropriate actions can be taken using the fault-tolerant controller to handle the fault in the system. In aircraft example, the supervisory module may decide to reconﬁgure the ﬂight controllers, the guidance system, and the navigation system based on the FTC type used in the system. There are also two types of FDD systems, namely passive FDD and active FDD systems. For the same example, the passive FDD systems “wait” until a fault or failure occurs, but in the active FDD systems, health-check ﬂying maneuvers or test signals injections via actuator’s commands can be used for health monitoring of the system. Different types of Kalman ﬁlter, for instance, can be used as a FDD block. However the focus of this thesis is mainly on fault-tolerant controllers but not FDD block.
Once the local supervisor automata are derived through the natural projection, the decentralized supervisor is then obtained using the parallel composition of local supervisor automata. Parallel composition captures the logical behavior of concurrent distributed systems by allowing each subsys- tem to evolve individually on its private events, while syn- chronize with its neighbors on shared events for cooperative tasks.
be one of the major weeds of cereal crops because this annual plant releases many seeds and develops resistance to a vari- ety of herbicides. Using a UAV, this species can be detected and mapped and its density can be estimated in winter wheat crops (Triticum aestivum L.) (Lambert et al. 2017). The authors acquired images in a range of 670–750 nm and RGB bands. In turn, Lu et al. ( 2017 ) investigated spatio-temporal variation of species composition in grassland. It is an essential step during the evaluation of grassland sensitivity to stress factors, the un- derstanding of evolutionary processes of the local ecosystem and the development of grassland management strategies. As an alternative to space-borne remote sensing images (e.g. MODIS, Landsat, Quickbird), the authors have used UAV with a modified digital camera (NIR, green and blue bands) to explore species composition in a tall grassland at different times of the growing season to assess spatio-temporal variations in species composition. Object-based plant classification was performed for the following classes: Bromus inermis (dense), B. inermis (sparse), Asclepias species, Solidago species, Festuca species with B. inermis and senesced grass. An interesting example of the UAV use is the mapping of aquatic vegetation. Husson et al. ( 2013 ) evaluated the use of a UAS for surveying emergent, floating-leaf and riparian vegetation in three boreal freshwater systems. Using RGB images acquired from UAV, the vegetation stands were visually and manually defined as homogenous patches that differed from surrounding vegetation patches in colour, texture and shape. Husson et al. ( 2016 ) also tested the possibility of using UAV images in the visible spectrum (380– 750 nm) for automatic mapping of helophytes (Equisetum fluvi- atile, Schoenoplectus lacustris, P. australis, Carex rostrata) and Nymphaeides (Nuphar lutea, Potamogeton natans, Spargan- ium spp., Nymphaea alba ssp. candida, Nuphar pumila). The automated classification of aquatic species was based on OBIA performed using two classification methods: threshold classifi- cation and random forest.
In 2009 Virginia Tech arranged a team of mechanical and aerospace engineering students to submit an entry for the International Aerial Robotics Competition (IARC). IARC charged competitors with navigating a UAV of their own throughout an indoor competition area with specific mission objectives and restrictions broadly elaborated in . In brief, the mission would include navigating the UAV through obstacles from point A to point B, collecting data once at point B, and returning it to point A where the mission began. Use of GPS, GLONASS, Galileo, or other satellite navigation systems was also strictly prohibited. Another critical rule of the competition was an all up mass limitation of 1.5 kg for all competing vehicles. This posed a challenge to Virginia Tech’s IARC team’s design since a critical sensor module weighing 536.1 grams assumed an overwhelming portion of their legalized mass. A solution was found in implementing lightweight yet strong carbon fiber with basswood frame in a quadrotor layout. Additionally, the team discovered benefits in placing the bulk mass right under the geometric center of the quadrotor as seen in Figure 2.3. Namely, improved flight stability along with a 360° sensor range of view were achieved. Further supporting this research, many subsequent image capture centered UAV platforms (DJI Series, Yuneec Q500) would go on to adopt layouts of this fashion.
Within Europe the rules that govern the flying of Unmanned Aircraft Systems, UAV, is dictated by European Aviation Safety Agency. Their current dictate is Regulation (EC) No 216/2008, mandates the Agency to regulate UAS and in particular Remotely Piloted Aircraft Systems, RPAS, when used for civil applications and with an operating mass of 150 Kg or more. Below this weight if not used commercially the individual member states define their own policy, and the training needed for the remote pilots does currently vary within the European Union, EU, and indeed they differ from the U.S.A. Federal Aviation Agency requirements. It is worth noting that individual states within the U.S.A. have particular specific requirements. Within the U.K. for aircraft of 20 kg or less, these are referred to as a 'small unmanned aircraft', which the requirements are a less stringent and are covered within Articles 166 and 167. Nevertheless, a 20kg object flying at speeds of 60 mph can cause harm, injury and death if not controlled. This paper ignores the requirements of pilots and concentrates only on maintenance. Nevertheless, a UAV still flying within the line of sight can crash and cause property damage, personal injuries or fatalities. A personal UAV will be bought with a warranty, not a maintenance contract and who is even going to be aware of what is acceptable for airworthiness or acceptable levels of safety .
Different multi-loop approaches have been utilized earlier. Few of them are: A static inner loop and an outer dynamic loop are designed using Eigen structure assignment and synthesis . Linear quadratic regulator technique and methods are employed for inner and outer loops, and four body angular measurements, rotor lag, flap state measurements and their derivatives are given as feedback . Likewise the other methods used are clearly explained in [12, 13, 14]. In spite of all this work, there are approaches which are theoretically attractive but have one or more of the issues like they are difficult to solve for the higher order systems, lack structure, difficult to implement.
In this chapter, we presented our obstacle avoidance algorithm for a teleoperated UAV. Based on a local obstacle state, created using a bin-occupancy filter with measurements from a depth camera and the robot’s state, the algorithm filters the operator’s input and alters it when necessary. The estimated obstacle state is used to predict possible collisions and to modify the velocity commanded by the operator to avoid obstacles. Additionally, we added an active avoidance component to compensate for any possible drift of the platform. Through the experiments presented in this chapter, we not only validated our navigation system but also the on-board state estimator and the self-sufficiency of the platform. The platform is able to estimate its state in an indoor, GPS-restricted environment, using IMU and optical flow integration and is independent from external tracking systems and computations. The initial outdoor testing shows that the algorithm is not limited to structured environments and can handle obstacles of different sizes.
There are many ideas to what makes an effective control strategy for swarms and/or UAVs. One that is simple and to the point is brought up in Parunak  and consists of four requirements, also referred to as the Four D’s. These are diverse, distributed, decentralized and dynamic. A control system must be diverse in its functions, the information it can handle, the entities that it can communicate with, and the sources that it can get information from. Distributed systems are important where there is any concern over issues with long-range communications bandwidth, by being distributed each component can just talk with its neighbors to pass messages along.
Since James Kuffner introduced the term “Cloud Robotics” in 2010, numerous studies have explored the benefits of this approach , . Cloud computing allows on-demand access to nearly unlimited computational resources, which is especially useful for bursty computational workloads that periodically require huge amounts of computation. Although the idea of taking advantage of remote computers in robotics is not new, the unparalleled scale and accessibility of modern clouds has opened up many otherwise unrealistic applications for mobile robot systems. For example, automated self-driving cars can access large-scale image and map data through the cloud without having to store or process this data locally . Cloud-based infrastructures can also allow robots to commu- nicate and collaborate with one another, as in the RoboEarth project .
A. Pedro Aguiar received the Licenciatura, M.S. and Ph.D. in Electrical and Computer Engineering from the Instituto Superior T´ecnico (IST), Technical University of Lisbon, Portugal in 1994, 1998 and 2002, respectively. Currently, Dr. Aguiar holds an Associate Professor position with the Department of Electrical and Computer Engineering (DEEC), Faculty of Engineering, University of Porto (FEUP). From 2002 to 2005, he was a post-doctoral re- searcher at the Center for Control, Dynamical- Systems, and Computation at the University of Cali- fornia, Santa Barbara (UCSB). From 2005 to 2012, he was a senior researcher with the Institute for Systems and Robotics at IST, and an invited assistant professor with the Department of Electrical and Computer Engineering, IST. His research interests include modeling, control, navigation, and guidance of autonomous robotic vehicles, nonlinear control, switched and hybrid systems, tracking, path-following, performance limitations, nonlinear observers, the integration of machine vision with feedback control, networked control, and coordinated/cooperative control of multiple autonomous robotic vehicles.
The ability to use low-cost multirotor platforms with autopilot systems provides distinct advantages for low altitude imagery platforms for intertidal reefs, including improved efficiency using pre-determined flight paths for data collection, critical to maximise survey time available at low tides. Off-the-shelf UAV platforms, easily accessible to the hobbyist, also provide the potential to enhance data output through citizen science programs, with low-cost platforms capable of autonomous programmed flight now common in the marketplace 12 , 28 . This readily available technology will allow citizen scientists access to pre-programmed flight paths, smart ground control targets to allow for cm precision, and web-based workflows to automate data upload to central deposi- tories for cloud processing and data dissemination. Over networks, this could provide scientists with high fre- quency image capture to monitor change not possible using scientific teams alone. Whilst yet to be fully exploited for citizen science, there is great potential for such UAV approaches to be effective in providing high frequency temporal data collection over targeted areas if adequate training and quality control can be provided, and flight regulatory requirements for small UAVs can be met 12 . There is also a need to quantify the detectability of species when using UAV surveys compared to traditional approaches to determine the value of low-cost UAVs to com- plement, or potentially replace, more field-intensive ground based approaches. Through accurate geo-referencing of imagery mosaics, UAVs provide the potential to identify subtle shifts in species distribution with repeat sur- veys. In addition, digital surface models from UAV surveys make it possible to collect data on the variation in geomorphic features, such as subtle changes in elevation and complexity that have been found to influence biotic assemblages 29 , and susceptibility to sea level rise with a changing climate 30 .
UAVs can overcome the problems in transporting sensors to the very position where an operator conducts direct VT or some other NDT method . Table 1. summarizes different aspects of the use of UAVs in RVT, as discussed in details previously in this section. The paper develops the concept of smart use of UAVs in RVT. Among the large variety of possible realizations of such a use, we assume a generic RVT in which an UAV carries a camera and a manipulator that includes an optical system. The novelty of our approach is that the manipulator is of modular structure, mounted to the UAV on the tested site just prior to flight. Having all stated in mind, the problem is how to design the manipulator so that is makes possible traversing significant portion of the vicinity of the hovering UAV that carries it. The software control of the manipulator dynamics is out of the scope of this article. Our approach is, thus, a part of the overall approach to enrich the autonomous vehicles with manipulators, which has been more developed in regard to ground-based autonomous vehicles , . Overall, the aim of that optically enriched manipulator is to enable the UAV camera to record the areas which are beyond its line of sight. The overall trajectory is complex because of the aforementioned aspects of the traversed region as well as because of the non-trivial, variable in time geometry of the UAVs with the manipulator mounted onto it.
The user community of unmannedaerial vehicle (UAV) systems has been growing significantly; the commercial market net worth reached a reported $8.3 billion in 2018 , the largest growth in the commercial markets being in small-class (un- der 55 pounds) vehicles. This paper develops an algorithm for a subset of this class of UAV, in particular, for fixed-wing vehicles. Users of small-class UAV are presently navigating autonomously by open-source algorithms such as Mission Planner, Cape, and Pix4D. These navigational systems employ waypoint naviga- tion (WN), wherein the user enters waypoints, whether a priori (static) or not How to cite this paper: Silverberg, L.M.
The performance of the system is calculated considering the sensor detection ranges and speed and the mean reaction time of a pilot. The work is continued in  with simulation of typical scenarios. Simulations show that the probability of the detection is 90% at the given detection range and that the probability of the collision avoidance is more than 85% in the presence of error. The main advantage of this system that it is scalable according to the requirements and the detected objects range information is available. Furthermore, the distance from the intruder can be detected is bigger compared to the EO sensor based systems. Also these systems can be used all time and all weather conditions. The main drawbacks are the size, weight, power consumption and relatively slow data rate (2 Hz).
SOð3Þ, shown in Fig 1 . However, it is important to highlight that the proposed monocu- lar SLAM method could be applied to other kind of platforms. The proposed method is mainly intended for local autonomous vehicle navigation. In this case, the local tangent frame is used as the navigation reference frame. Thus, the initial position of the vehicle defines the origin of the navigation coordinates frame. The navigation system follows the NED (North, East, Down) convention. The magnitudes expressed in the navigation and in the camera frame are denoted respectively by the superscripts N and C . All the coordinate systems are right-handed defined. It is also assumed that the location of the origin of camera frame respect to other ele- ments of the quadcopter (e.g. GPS antenna) is known and fixed. In this case, the position of the origin of the vehicle can be computed from the estimated location of the camera.
presented applications have specific drawbacks that should be taken into account. That is, the vision-based systems are low cost sensor devices, which provides high amount of information, but have the drawback of the high sensitivity to lighting conditions (e.g. direct sun light may lead to lack of information). Moreover, all the presented algorithms and applications give full understanding and convergence to
There are several planning strategies proposed for ground robots , delivery systems , autonomous high-speed, fixed- wing UAV networks , or mobile sensor networks  with different objectives and constraints. Applications range from snow removal, lawn mowing, floor cleaning, to surveillance, mobile target tracking, chemical or hazardous material detec- tion and containment, or to any combination of localization and navigation problems (see , , ). While some algorithms use prior information and have exact or partial decomposition of the areas, others use sensor-based informa- tion in unknown environments to make navigation decisions. Algorithms exist that try to minimize the path traveled or time or energy required to achieve a goal. These different schemes have some common building blocks, such as static or dynamic area decomposition, cooperative or non-cooperative actions, individual or collective decisions, static or adaptive behavior. Therefore, in this paper, we study two different approaches that consider some intuitive combination of these building blocks. The remainder of the paper is organized as follows. The system model and metrics of interest are given in Section II. The methodologies are introduced in Section III. Results are given in Section IV and the paper is concluded in Section V.