The quadrotor helicopter is a unique flying vehicle which uses the thrust from four motors to provide hover flight capability. The uncoupled nature of the longitudinal and lateral axes and its ability to support large payloads with respect to its size make it an attractive vehicle for autonomous vehicle research. In this thesis, the quadrotor is modeled based on first principles and a proportional-derivative control method is applied for attitude stabilization and position control. A unique means of using an optic flow sensor for velocity and position estimation in an indoor setting is presented with flight results. Reliable hover flight and hallway following capabilities are exhibited in GPS-denied indoor flight using only onboard sensors. Attitude angles can be reliably estimated in the short run by integrating the angular rates from MEMS gyros, but noise on the signal leads to drift which renders the measurement unsuitable to attitude estimation. Typical methods of providing vector attitude corrections such as accelerometers and magnetometers have inherent weaknesses on hovering vehicles. Thus, an additional vector measurement is necessary to correct attitude readings for long-term flights. Two methods of using image processing to determine vanishing points in a hallway are demonstrated. The more promising of the two uses a Hough transform to detect lines in the image and forms a histogram of the intersections to detect likely vanishing point candidates. Once the vanishing point is detected, it acts as a vector measurement to correct attitude estimates on the quadrotor vehicle. Results using onboard vision to estimate heading are demonstrated on a test stand. Together, these capabilities improve the utility of the quadrotor platform for flight without the need of any external sensing capability.

Vertical Take-Off and Land (VTOL) Unmanned Air Vehicles (UAVs) possess several desirable characteristics, such as being able to hover and take-off or land in confined areas. One type of VTOL airframe, the tailsitter, has all of these advantages, as well as being able to fly in the more energy-efficient level flight mode. The tailsitter can track trajectories that successfully transition between hover and level flight modes. Three methods for performing transitions are described: a simple controller, a feedback linearization controller, and an adaptive controller. An autopilot navigational state machine with appropriate transitioning between level and hover waypoints is also presented. The simple controller is useful for performing a immediate transition. It is very quick to react and maintains altitude during the maneuver, but tracking is not performed in the lateral direction. The feedback linearization controller and adaptive controller both perform equally well at tracking transition trajectories in lateral and longitudinal directions, but the adaptive controller requires knowledge of far fewer parameters.

Utilizing the Euler-Rodrigues symmetric parameters (attitude quaternion) to describe vehicle orientation, we develop a multiplicative, nonlinear (extended) variation of the Kalman filter (MEKF) to fuse data from low-cost sensors. The sensor suite is comprised of gyroscopes, accelerometers, and a GPS receiver. In contrast to the common approach of using the complete vehicle attitude as the quantities to be estimated, our filter states consist of the three components of an attitude error vector. In parallel with the time update of the attitude error estimate, we utilize the gyroscope measurements for the time propagation of the attitude quaternion. The accelerometer and the GPS sensors are used independently for the measurement update portion of the Kalman filter. For both sensors, a vector arithmetic approach is used to determine the attitude error vector. Following each measurement update, a multiplicative reset operation moves the attitude error information from the filter state into the attitude estimate. This reset operation utilizes quaternion algebra to implicitly maintain the unity-norm constraint. We demonstrate the effectiveness of our attitude estimation algorithm through flight simulations and flight tests of aggressive maneuvers such as loops and small-radius circles. We implement an approach to aerobatic maneuvering for miniature air vehicles (MAVs) using time-parameterized attitude trajectory generation and an associated attitude tracking control law. We designed two methodologies, polynomial and trigonometric, for creating functions that specify pitch and roll angles as a function of time. For both approaches, the functions are constrained by the maneuver boundary conditions of aircraft position and velocity. We construct a trajectory tracking feedback control law to regulate aircraft orientation throughout the maneuvers. The trajectory generation algorithm was used to construct several maneuvers and trajectory tracking control law successfully executed the maneuvers in the flight simulator. In addition to the simulation results, MAV flight tests verified the performance of the maneuver generation and control. To achieve obstacle avoidance maneuvering, the time parameterized trajectories were converted to spatially parameterized paths, which allowed for inertial reference frame position error to be included in the control law feedback loop. We develop a novel method to achieve the spatial parameterization using a prediction and correction approach. Additionally, the first derivative of position of the desired path is modified using a corrective parameter scheme prior to being used in the control. Using the path position error and the corrected derivative, we utilize a unit-norm quaternion framework to implement a proportional-derivative (PD) control law. This control law was demonstrated in simulation and hardware on maneuvers designed specifically to avoid obstacles, namely the Immelmann and the Close-Q, as well as a basic loop.

The viability of small Unmanned Aerial Vehicles (UAVs) as a stable platform for specific application use has been significantly advanced in recent years. Initial focus of lightweight UAV development was to create a craft capable of stable and controllable flight. This is largely a solved problem. Currently, the field has progressed to the point that unmanned aircraft can be carried in a backpack, launched by hand, weigh only a few pounds and be capable of navigating through unrestricted airspace. The most basic use of a UAV is to visually observe the environment and use that information to influence decision making. Previous attempts at using visual information to control a small UAV used an off-board approach where the video stream from an onboard camera was transmitted down to a ground station for processing and decision making. These attempts achieved limited results as the two-way transmission time introduced unacceptable amounts of latency into time-sensitive control algorithms. Onboard image processing offers a low-latency solution that will avoid the negative effects of two-way communication to a ground station. The first part of this thesis will show that onboard visual processing is capable of meeting the real-time control demands of an autonomous vehicle, which will also include the evaluation of potential onboard computing platforms. FPGA-based image processing will be shown to be the ideal technology for lightweight unmanned aircraft. The second part of this thesis will focus on the exact onboard vision system implementation for two proof-of-concept applications. The first application describes the use of machine vision algorithms to locate and track a target landing site for a UAV. GPS guidance was insufficient for this task. A vision system was utilized to localize the target site during approach and provide course correction updates to the UAV. The second application describes a feature detection and tracking sub-system that can be used in higher level application algorithms.

This thesis presents the design of an autopilot capable of flying small unmanned aerial vehicles with wingspans less then 21 inches. The autopilot is extremely small and lightweight allowing it to fit in aircraft of this size. The autopilot features an advanced, highly autonomous flight control system with auto-launch and auto-landing algorithms. These features allow the autopilot to be operated by a wide spectrum of skilled and unskilled users. Innovative control techniques implemented in software, coupled with light weight, robust, and inexpensive hardware components were used in the design of the autopilot.

Using autonomous miniature air vehicles (MAVs) is a cost-effective, simple method for collecting data about the size, shape, and location characteristics of a forest fire. However, noise in measurements used to compute pose (location and attitude) of the on-board camera leads to significant errors in the processing of collected video data. Typical methods using MAVs to track fires attempt to find single geolocation estimates and filter that estimate with subsequent observations. While this is an effective method of resolving the noise to achieve a better geolocation estimate, it reduces a fire to a single point or small set of points. A georeferenced mosaic is a more effective method for presenting information about a fire to fire fighters. It provides a means of presenting size, shape, and geolocation information simultaneously. We describe a novel technique to account for uncertainty in pose estimation of the camera by converting it to the image domain. We also introduce a new concept, a Georeferenced Uncertainty Mosaic (GUM), in which we utilize a Sequential Monte Carlo method (a particle filter) to resolve that uncertainty and construct a georeferenced mosaic that simultaneously shows size, shape, geolocation, and uncertainty information about the fire.

Unmanned vehicle systems, specifically unmanned air vehicles (UAVs) and unmanned ground vehicles (UGVs), have become a popular research topic. This thesis discusses the potential of a UAV-UGV system used to track a human moving through complex urban terrain. This research focuses on path planning problems for both a UAV and a UGV, and presents effective solutions for both problems. In the UAV path planning problem, we desire to plan a path for a miniature fixed-wing UAV to fly through known urban terrain without colliding with any buildings. We present the Waypoint RRT (WRRT) algorithm, which accounts for UAV dynamics while planning a flyable, collision-free waypoint path for a UAV in urban terrain. Results show that this method is fast and robust, and is able to plan paths in difficult urban environments and other terrain maps as well. Simulation and hardware tests demonstrate that these paths are indeed flyable by a UAV. The UGV path planning problem focuses on planning a path to capture a moving target in an urban grid. We discuss using a target motion model based on Markov chains to predict future target locations. We then introduce the Capture and Propagate algorithm, which uses this target motion model to determine the probabilities of capturing the target in various numbers of steps and with various initial UGV moves. By applying some different cost functions, the result of this algorithm is used to choose an optimal first step for the UGV. Results demonstrate that this algorithm is at least as effective as planning a path directly to the current location of the target, and that in many cases, this algorithm performs better. We discuss these cases and verify them with simulation results.