A Vision-based Method For Solving Heading Attitude Of Airdrop Cargo Platform

Airborne airdrop technology is the foundation of the Air Force’s equipment development.In order to make long-distance airborne equipment land smoothly,it is necessary to obtain the attitude data of the cargo platform module to adjust the landing attitude in real time to avoid landing rollover accidents.At present,traditional attitude measurement equipment will accumulate errors,so it is urgent to study a system for measuring the attitude of airdrop cargo platforms.Because visual navigation has the advantages of low cost and strong concealment,this paper proposes a research on the attitude calculation technology of airdrop cargo platforms based on visual recognition of ground cooperation targets and line features.(1)Aiming at the airdrop scene with weak texture,design the overall cooperation goal of "house shape" as a visual sign on the ground,and propose an airdrop cargo platform attitude calculation method based on the cooperation goal.This method divides the landing distance of the airdrop system into two stages,far-ground and nearground,designing overall cooperation goals,designing different scale signs for different landing stages as a visual reference,and effectively increasing the autonomous landing height of the airdrop cargo platform.In the far-ground stage,the inherent geometric properties of the cooperative target are used to identify in the airdrop scene;in the near-ground stage,the central cooperative target is screened through the Hu invariant moment as a ground auxiliary feature,and the homography is finally used to carry out the movement of the airdrop cargo platform.(2)Aiming at the line segment features such as the ground texture and the edge of the river in the airdrop scene,an algorithm for calculating the posture of the airdrop cargo platform based on the line feature is designed.The line feature is more stable than the point feature in the cooperative target,and there is no need to track the cooperative target.According to the LSD line feature extraction algorithm,there is no mechanism for selecting the length of short line segments and merging short line segments into long line segments.This paper uses the angle characteristics,spatial position characteristics and length characteristics of each line segment to group and merge broken lines,and remove the local dense line features to solve the disadvantages of LSD algorithm oversegmentation.Then use the improved RANSAC algorithm to remove the mismatch of line segments,improve the real-time performance of the algorithm and the accuracy of the calculated attitude angle,and make up for the problem of difficult extraction of point features or low matching accuracy in the airdrop environment.Finally,a constraint equation is established for the line features that have been successfully matched,and the homography is used to obtain the three-axis attitude angle during the landing of the airdrop cargo platform.Finally,the performance of the algorithm is verified by UAV to simulate the trajectory of the landing of the airdrop cargo platform.Experiments show that extracting cooperative targets or line features in the airdrop scenario can meet the requirements of attitude calculation of airdrop cargo platform.Experiments show the significance and feasibility of the vision-based airdrop cargo platform heading and attitude calculation technology,and it has good application prospects.