Study On Vision-Based Measurement Of Dynamic Motion For Aircraft Model Suspended By Wire System

The Wire-Driven Parallel Robot(WDPR)provides a new supporting method for wind tunnel tests.It has the advantages of strong adaptability,large motion space and low interference and can be used for cornplex and dynamic multi/six-DOF motion in wind tunnel tests.The high-precision and real-time measurement for the position and attitude of the aircraft model suspended by WDPR is especially critical for achieving closed-loop control of the system.The traditional measurement methods of wind tunnel model position and attitude has a limited effect,and it is difficult to meet the require-ments of wind tunnel dynamic tests.Therefore,it is necessary to study the dynamic measurement of the position and attitude for the aircraft model suspended by WDPR.Photogrammetry is very suitable for wind tunnel model position and attitude meas-urement because of its non-contact,high precision and low interference.Aiming at the existing problems of dynamic pose measurement based on monocular vision,the dy-namic pose measurement methods based on RGB-D camera and binocular stereo vision are studied.The main work is shown belowFirstly,aiming at the vision-based position and attitude measurement of aircraft model suspended by WDPR,the relevant measurement coordinate system is con-structed and the camera pinhole imaging model is mathematically modeled.The cali-bration of the internal and external parameters of the pinhole camera and the relative pose of the binocular camera are realized.Secondly,a dynamic measurement method for model position and attitude based on RGB-D camera is developed.ICP algorithm is used to estimate initial position and attitude value through the extraction of SIFT feature point for RGB image,correct match and the reconstruction of 3D.Then a unconstrained least squares optimization problem under Lie algebra is constructed based on re-prOojection error and optimized by LM algorithm.The stability analysis of the measurement system and the static experi-mental validation is conducted.Finally,a dynamic measurement method for model position and attitude based on binocular vision is developed.A kind of encoded round marker is designed and can be affixed on the model surface.The interference of cables to markers imaging is elimi-nated by image processing,following by the reconstruction of markers'3D coordinates.Then the initial values of relative position and attitude are estimated by Absolute Ori-entation algorithm and theoretical error analysis is given,and the optimal result is achieved through building an unconstrained least squares optimization problem in Lie Algebra based on re-projection error,which is solved by L-M optimization algorithm.The stability analysis of the measurement system is carried out.The static and dynamic experimental validations and three typical dynamic motions for aircraft model are re-spectively conducted,such as heave/pitch oscillation,and the factors affecting the ac-curacy of dynamic measurement are qualitatively analyzed.