Advances in Unmanned Aerial Vehicles

This Chapter justifies the rationale for publishing this edited book. It starts with a non technical, general discussion about unmanned aerial vehicles (UAVs). Then, it presents some fundamental definitions related to UAVs for clarification purposes, and discusses the contents of the book in a very concise way. It paves the way for what is included in subsequent Chapters and how the material, even though it is divided in parts, ties together in a rather unified and smooth way. The goal is to help the potential reader be-come familiar with the contents of the book and with what to expect reading each Chapter.

This ‘pictorial’ Chapter presents a historical perspective on unmanned aerial vehicles (UAVs) starting from Ancient Greece to the beginning of the 21^st Century. The UAV history, from a very early dream to today’s reality is illustrated through a series of figures with detailed legends that are arranged mostly chronologically; they reveal the unmanned vehicle evolution and designs over a period of almost 2,500 years. The Chapter, even though it is non-technical, offers an accurate glimpse of history and helps the reader understand the tremendous level of growth in the unmanned systems area. Almost all figures have been taken from archives and web sites available on-line. The list is by no means complete, but it is very informative. The Chapter layout and contents are similar to Chapter 1 of reference [10].

The goal of this Chapter is to present fundamental background information related to the derivation of the basic equations of motion of a traditional airplane, explain how the airplane’s position and orientation are determined with respect to a reference frame (Earth-fixed inertia reference frame), derive the aerodynamic forces that act on the airplane, define the corresponding control angles, and conclude with derivation of the open-loop dynamics. The material included in this Chapter is a very concise version of what may be found in any related textbook, and follows the same notation and derivation approach described in the references.

Miniature helicopters are increasingly used in military and civilian applications, mainly due to their ability to hover, fly in very low altitudes and within confined spaces. However, due to model nonlinearities and inherent instabilities, low-level controller design for autonomous flights is a challenge. This Chapter presents an overview of major accomplishments in the area of unmanned helicopter control by several research groups, and focuses on techniques used for low-level control. It then describes a general model suitable for small or miniature helicopter non-aggressive flights and compares three different controllers, a PID, a Linear Quadratic Regulator (LQR) and an H ₈ controller in terms of their practical implementation to achieve autonomous, self-governing flights.

Unmanned helicopters are being widely used in a multitude of applications, mostly military, as well as civilian. Their usefulness and applicability results from their ability to fly in very low altitudes, hover, cruise, take off and land almost everywhere, executing missions that require aggressive and non-aggressive flight scenarios.

Miniature aerial vehicles (MAVs) have attracted major research interest during the last decade. Recent advances in low power processors, miniature sensors and control theory have contributed to system miniaturization and creation of new application fields.

Unmanned aerial vehicles (UAVs) are playing increasingly prominent roles in defense programs and strategy around the world. Technology advancements have enabled the development of large UAVs (e.g., Global Hawk, Predator) and the creation of smaller, increasingly capable UAVs. The focus of this Chapter is on smaller fixed-wing miniature aerial vehicles (MAVs), which range in size from % to 2 m in wingspan. As recent conflicts have demonstrated, there are numerous military applications for MAVs including reconnaissance, surveillance, battle damage assessment, and communications relays.

The human pilot in manned aircraft acts as a closed-loop control system by taking in visual data in addition to sensor measurements and then responding to received data. For example, the pilot may observe the changing surroundings to better understand the movements of his aircraft, or the pilot can direct the aircraft to a region of interest after taking ground observations. In unmanned aircraft, the absence of a pilot requires development of automated control systems.

The localization problem is usually solved in aerial robotics by means of GPS, Inertial Measurement Units (IMUs) and compass sensors, which are fused to provide UAV position estimation in a global frame. However, GPS estimates are often subject to inaccuracies and errors related to GPS failure or degradation. Also, GPS cannot be directly used for relative positioning of a UAV with respect to other objects, as for example for landing on mobile platforms or for wall following in building inspection.

Path planning refers to the generation of a space path between an initial location and the desired destination, with an optimal or near-optimal performance under specific constraints [1]. A path planning algorithm may produce different candidate plans, which should be compared and evaluated based on specific criteria. Such criteria are generally related to feasibility and optimality of the path generation. The first criterion relates to derivation of a plan that moves safely a UAV (an object) to its final state, without taking into account the quality of the produced plan. The second criterion refers to derivation of optimal, yet feasible, paths, with optimality defined according to the problem under consideration [2]. However, searching for optimal paths is not a trivial task; in most cases this requires excessive computational time; in some cases even computation of just one feasible path is a rather involved task. Therefore, the search focuses mostly on suboptimal or just feasible solutions.

The last decade has seen an intense world wide effort in developing mini aerial vehicles (mAVs). Such vehicles are characterized by small scale (dimensions of the order of 60cm), limited pay load capacity and embedded avionics systems. A key component of the avionics system in a mAV is the attitude estimation subsystem [2] 12] [30]. Such systems must be highly reliable and have low computational overhead to avoid overloading the limited computational resources available in some applications. Traditional linear and extended Kaiman filter (EKF) techniques [14] [3] [20] suffer from issues associated with poor system modeling (in particular characterization of noise within the system necessary for tuning filter parameters) as well as potentially high computational requirements [28] [30]. An alternative is to use deterministic complementary filter and nonlinear observer design techniques [35] [2] [1] [34]. Recent work has focused on some of the practical issues encountered when data is obtained from low cost inertial measurement units (IMU) [26] [1] [34] [23] as well as observer design for partial attitude estimation [27] [21] [22]. It is also worth mentioning the related problem of fusing IMU and vision data [16] [25] [13] [7] [6] and the problem of fusing IMU and GPS data [24] [34]. A key issue in attitude observer design for systems with low-cost IMU sensor units is on-line identification of gyro bias terms. This problem is also important in IMU calibration of attitude observers for satellites [14] [8] [4] [32] [17].

Development of a solar powered aircraft capable of continuous flight was still a dream a few years ago; this great challenge has become reality today. Significant progress has been made in the domains of flexible solar cells, high energy density batteries, miniaturized MEMS and CMOS sensors and powerful processors.

Homeland security and disaster mitigation efforts are often taken place in unforeseen environments that include caves, tunnels, forests, cities, and even inside urban structures. Performing various tasks such as surveillance, reconnaissance, bomb damage assessment or search and rescue within an unfamiliar territory is not only dangerous but it also requires a large, diverse task force. Unmanned robotic vehicles could assist in such missions by providing situational awareness without risking the lives of soldiers, first responders, or other personnel. While ground-based robots have had many successes in search and rescue situations [6], they move slowly, have trouble traversing rugged terrain, and can still put the operator at risk. Alternatively, small unmanned aerial vehicles (UAVs) can provide soldiers and emergency response personnel with an “eye in the sky” perspective. On an even smaller scale, tiny bird-sized aircraft or micro air vehicles (MAVs) can be designed to fit in a backpack and can be rapidly deployed to provide surveillance and reconnaissance in and around buildings, caves, tunnels and other near-Earth environments. Navigating in these environments, however, remains a challenging problem for UAVs. In [7], promising results are shown for a rotorcraft equipped with a SICK laser scanner. However, because lift decreases with platform size, carrying this type of sensor on a MAV is not feasible.

The role of autonomous surveillance has proven to be important and applicable to a wide range of applications such as target location, map building, border security, pollution monitoring and control, and battle damage assessment. UAVs fit into the scenario of autonomous surveillance perfectly as they involve a low risk factor and facilitate technological advancements, making their use feasible in real world scenarios. UAVs are generally classified by their flight altitude, launch and recovery methods as detailed [23].

Control problems involving unmanned mobile vehicles have attracted considerable attention in the control community during the past decade. One of the basic motion tasks assigned to a mobile vehicle may be formulated as following a given trajectory [13] [25]. The trajectory tracking problem was globally solved in [20] by using a time-varying continuous feedback law, and in [2] [12] [16] through the use of dynamic feedback linearization. The backstepping technique for trajectory tracking of nonholonomic systems in chained form was developed in [6] [10]. In the special case when the vehicle model has a cascaded structure, the higher dimensional problem can be decomposed into several lower dimensional problems that are easier to solve [17].

Significant advances have been made in recent years in volcanic eruption forecasting and in understanding the behaviour of volcanoes. A major requirement is improvement in the collection of field data using innovative methodologies and sensors. Collected data are typically used as input for computer simulations of volcanic activity, to improve forecasts for longlived volcanic phenomena, such as lava flow eruptions and sand-rain.

This Chapter has been motivated by the challenge to design and implement an on-board processing system that meets very limited payload capabilities of small unmanned aerial and ground vehicles while sacrificing minimal computational power and run time, still adhering to the low cost nature of commercial Radio Controlled (RC) equipment. Fundamental issues justifying the implementation of such an on-board system are increased autonomy, increased processing throughput, diversity in areas of application, and increased safety.