Research On Video-Based UAV Formation Flight Control

Video-based UAV formation flight control is a new intelligent processing technology of UAV application development in future battlefield, which is used to solve the problems, such as manual monitoring is expensive, investigators perform tasks are at high risk, people cannot reach the place of mission requirements, and over enemy combatants need radio silence.Considering the disadvantages of a single UAV when performing tasks, multiple-UAV coordinated formation flight based on UAV vision system is introduced, and the “1” formation,which belongs to "Leader-Follower" formation, is studied as an example. Starting from the structure and function of the visual system to create a UAV vision system, a CCD camera and a video capture card are used to capture and preprocess digital image information of the lead aircraft, and the camera motion equations are built by using coordinate transformation.A three-time difference based background subtraction technique is proposed, which is used for processing the acquired image sequence of the lead aircraft and providing target tracking position. At the same time, a Mean-shift tracking algorithm for UAV based on TWH and FB-error constraint is proposed. A new TWH is introduced to effectively separate the target and its background, which is similar with itself, by improving weights of the pixels.Based on the forward-backward consistency assumption in which correct tracking should be reversible over the time flow, a new concept called FB-error is proposed. A forward-backward evaluation scheme has been put forward by using FB-error constraint in tracking. This algorithm made the Mean-shift algorithm be more robust to short-term occlusion problem and solve the tracking failure when UAV fuselage is similar to clouds or sheltered from cloud cluster.Based on camera motion equations, the lateral deviation angle ? and longitudinal deviation angle ? between the two aircraft are calculated when the "1" formation is changed by reason of the change in lead aircraft heading maneuvering, then the relative distance between the two aircraft are estimated, and three channels PID algorithm formation flyingcontrol using the relative distance as input is achieved. Simulation results show that the UAV vision system can effectively assist wingman in tracking the lead aircraft flying in formation, and the formation can be timely corr ected and restored to the desired formation again.