This research proposed the use of inexpensive, lightweight range sensors for indoor unmanned aerial vehicle (UAV) navigation. Two potential range sensors were tested for suitability and error characteristics. The SHARP infrared range sensors provide a narrow beam and a higher resolution distance measurement, at the expense of de creased range (approximately 150–180 cm maximum). The MaxBotix® EX1™ ultrasonic range sensors had a longer range (up to 6.45 m) and a wider beam.
In addition, the sonar has unique measurement properties due to the physics of the metrology technique used for ranging. The sonar sends out a sound pulse that expands on a spherical front. When this wavefront encounters an object, it is reflected back toward the sensor. When the first return is detected, a time-of-flight calculation is used to determine range. As a result, the perpendicular distance to a wall will always have the shortest return. Hence, if one of the sonar can detect a wall at all, it will always return the perpendicular distance to that wall.
Similar effects were observed when the sensors were aimed at the corner of a room. Thus, the sonar range sensors return the same distance regardless of moderate changes in angle to the target. This makes them useful for altitude ranging, as the vehicle roll and pitch do not greatly affect the distance measurement to the ground. The sonar are less useful for scanning a room to make a 2D plot of the room, although by using a histogram analysis, room dimensions were determined from experimental sonar data. In addition, simulations were performed using sonar and the histogram analysis to determine room dimensions in real time. The addition of a yaw-rate gyro was simulated, which, along with wall-following behavior to correct for drift, could allow a heading estimate to be maintained during the flight as well. With a heading estimate and the combination of sonar and IR sensors for ranging, a small, passively stable aerial platform could be used for basic mapping and localization.
Once the range sensors were analyzed, and their measurement and error characteristics were determined, simulations were completed to develop and refine mapping and localization techniques using the low-cost range sensors. Finally, a two-phase flight test program was initiated to further develop navigation, guidance, and control algorithms. During the first phase of the flight test, altitude control and longitudinal control were demonstrated. During the second phase of the flight test program, heading control and lateral control were developed. The completion of the flight test program demonstrated the capability to navigate unknown indoor environments using only simple, ultra-low-cost range sensors mounted to a passively stable coaxial rotorcraft vehicle.
This work was done by D. Michael Sobers, Jr., Girish Chodhary, and Eric N. Johnson of Georgia Institute of Technology. For more information, download the Technical Support Package (free white paper) at www.defensetechbriefs.com/tsp under the Physical Sciences category. GIT-0001
This Brief includes a Technical Support Package (TSP).
Indoor Navigation for Unmanned Aerial Vehicles
(reference GIT-0001) is currently available for download from the TSP library.
Don't have an account? Sign up here.