UAV Landing Based on the Optical Flow Videonavigation

Abstract

An automatic landing of an unmanned aerial vehicle (UAV) is a non-trivial task requiringa solution of a variety of technical and computational problems. The most important is the precisedetermination of altitude, especially at the final stage of approaching to the earth. With currentaltimeters, the magnitude of measurement errors at the final phase of the descent may be unacceptablyhigh for constructing an algorithm for controlling the landing manoeuvre. Therefore, it is desirableto have an additional sensor, which makes possible to estimate the height above the surface of therunway. It is possible to estimate all linear and angular UAV velocities simultaneously with thehelp of so-called optical flow (OF), determined by the sequence of images recorded by an onboardcamera, however in pixel scale. To transform them into the real metrical values it is necessary toknow the current flight altitude and the camera angular position values. The critical feature of theOF is its susceptibility to the camera resolution and the shift rate of the observed scene. During thedescent phase of flight, these parameters change at least one hundred times together with the altitude.Therefore, for reliable application of the OF one needs to coordinate the shooting parameters withthe current altitude. However, in case of the altimeter fault presence, the altitude is also still to beestimated with the aid of the OF, so one needs to have another tool for the camera control. One of thepossible and straightforward ways is the camera resolution change by pixels averaging in computerpart which performed in coordination with theoretically estimated and measured OF velocity. Thearticle presents results of such algorithms testing from real video sequences obtained in flights withdifferent approaches to the runway with simultaneous recording of telemetry and video data.

Embodiment and Its Influence on Informational Costs of Decision Density—Atomic Actions vs. Scripted Sequences

“…In this case, the sensor can be seen as implicitly activating the "same" action for as long as the stimulus exists. With nature setting the example, this technique has been adapted to be used in artificial unmanned aerial vehicles (Miller et al, 2019). Of course, the organism needs the ability to sense the gradient or a suitable proxy for it to generate such a label which obviously is easier for some tasks than for others.…”

An optical-flow-based monitoring method for measuring translational motion in infrared-thermographic images of AFP processes

“…Consequently this analysis allows the positioning of images relative to each other by extracting suitable features that can be also recognized in the next image. Miller [8] uses optical-flow analysis for video navigation of unmanned aircraft by using the algorithm of Lucas [9] to determine motion vectors of local image areas in successive images. Nooralishahi [10] uses optical-flow analysis with Lowe's Scale-Invariant Feature Transform (SIFT) method [11] to stitch together aerial thermographic images of the earth's surface.…”

UAV Flight and Landing Guidance System for Emergency Situations †

“…Our proposed safe landing system can avoid the obstacles in all directions based on optical flow, and can effectively cope with the complex environment and various obstacles in the urban area. Miller et al estimated reliable altitude using the difference between the optical flow velocity and calculated via exact formulas [26]. Estimation of altitude is tested from video sequences obtained in flights, but altitude is not calculated during the actual flight.…”