Capt Dayvid Prahl recently graduated from the Air Force Institute of Technology where he performed research for the Advanced Navigation Technology Center. The Air Force as a need for precision navigation; however, the systems that the Air Force uses—especially unmanned systems—keep getting smaller and therefore will be able to operate in places airplanes have never been able to fly before, such as inside buildings, down long urban canyons, in dense urban areas, or even underground. In these environments, GPS is ineffective because satellite signals are blocked entirely or highly degraded. Since GPS is unavailable, an alternate technology is needed to accomplish the navigation task. Capt Prahl’s research sought to combine an inertial sensor and a small webcam to obtain a precise navigation solution in an indoor corridor. Inertial navigation sensors are one technology that can be used instead of GPS to obtain a navigation solution. Inertial sensors consist of accelerometers and gyroscopes and they provide a good solution in the short-term. In order to use an inertial sensor in the long-term, one would need to augment it with information from a second sensor. For Capt Prahl’s research, the second sensor is the camera. The quadrotor is an example of a small unmanned vehicle that can fly indoors. It’s called a quadrotor because of the four counter rotating propellers in each corner of the vehicle. It’s controlled by adjusting the speed at which the propellers spin. By turning up propellers on the same side of the vehicle, it can be turned one way or the other. It’s equipped with a small inertial sensor, which is a lightweight, low power sensor that also has a very fast drift, meaning it cannot be used for very long without being updated with additional information from another sensor. The quadrotor’s camera provides the additional information that the inertial sensor needs to keep a stable altitude in position determination. Capt Prahl’s research was based on one fundamental assumption—that the vehicle is operating inside a man-made environment. This environment is assumed to consist of many planar surfaces with vertical walls and flat, level ceilings and floors that will meet at right angles. The corners where the ceilings and walls meet will form lines that are parallel, but in an image of the environment, appear to converge at one point—the vanishing point. By tracking the vanishing point, one can determine the orientation of the camera with respect to the scene. By knowing the orientation of the camera, the orientation of the vehicle can also be determined.
