Research On Cabin Displacement Detection Technology Under Automatic Leveling Mode Of Boarding Bridge

After the boarding bridge is docked with the cabin,the height of the aircraft varies with the load.Therefore,the boarding bridge must work in the automatic leveling mode to avoid collision with the cabin.The boarding bridge achieves this function through the encoder on the leveling wheel,but the mechanism will have false contact,skid and other problems,and the boarding bridge operator needs to check after each docking.At present,the research of unmanned boarding bridge not only replaces the bridge hand,but also introduces a new research topic,that is,how to detect cabin displacement more accurately and intelligently.In this paper,the marker-free displacement measurement method based on vision is analyzed,and the cabin displacement detection technology based on sparse optical flow and feature point matching is mainly studied.The main research contents of this paper are as follows:First of all,considering that the relative motion of cabin and the boarding bridge only shows the characteristics of translational transformation in the imaging plane,a weighted displacement calculation method based on translational transformation is proposed.The Improved Progressive Sample Consensus method was used to calculate the transformation model and remove the outer points that did not conform to the model.An threshold,distinguish the threshold values of the inner and outer points,was designed to depends on sample set.Finally,the weighted method was used to calculate the mean displacement of each inner point.Secondly,in view of the high demand of feature matching method on the speed of feature extraction algorithm,according to the characteristics that the engine room to be detected in this paper has simple surface texture and a large number of patches of color invariant areas,a feature point extraction algorithm based on block is proposed.By setting two parameters,block size and image block standard deviation threshold,the image block whose standard deviation is less than the threshold value is judged to have no feature point.At the same time,the design parameters can adapt to the scale changes of the pyramid layer to speed up the calculation speed and achieve the effect of multi-scale block size setting more effectively.Thirdly,according to the characteristics of fast computation speed and large cumulative error of optical flow method,a displacement detection algorithm fusing optical flow and matching method is designed in this paper.The displacement is measured by the optical flow method at the non-key frame,and the initial tracking results of the optical flow method and the matching method are fused to calculate the target displacement at the key frame.Finally,experiments are designed to verify that the new displacement calculation method adopted in this paper can improve the performance of displacement measurement,and the improvement of the matching method can accelerate the detection speed of feature points and improve the illumination robustness of the algorithm.Compared with the sparse optical flow method,the fusion algorithm can effectively reduce the cumulative error and meet the needs of long-term displacement detection,although the fusion algorithm has a slightly lower speed on the cabin measurement.