Velocity Estimation for UAVs by Using High-Speed Vision

In recent years, applications of high-speed visual systems have been well developed because of their high environmental recognition ability. These system help to improve the maneuverability of unmanned aerial vehicles (UAVs). Thus, we herein propose a high-speed visual unit for UAVs. The unit is lightweight and compact, consisting of a 500 Hz high-speed camera and a graphic processing unit. We also propose an improved UAV velocity estimation algorithm using optical flows and a continuous homography constraint. By using the high-frequency sampling rate of the high-speed vision unit, the estimation accuracy is improved. The operation of our high-speed visual unit is verified in the experiments.