| Real-time attitude information of the carrier(i.e.,aircraft,missiles,and vehicles etc)is very important for the control system to judge the movement state of the ca rrier.The traditional attitude determination technology mostly adopts the GPS(Global Positioning System)multi-antenna system,which is high-precision while high-cost and complex to realize(i.e.,the cycle-slip detection and integer ambiguity resolution etc).Furthermore,the accuracy and reliability of GPS multi-antenna system is susceptible to the surrounding observation environment.Compared to multi-antenna system,GPS single-antenna is simple to erect and.relatively low-cost Since GPS signals are easily locked or interrupted in urban canyons and tunnels,this contribution attempts to integrate GPS single-antenna system,INS(Inertial Navigation System),and visual system to obtain the reliable carrier attitude information,which can not only overcome the weaknesses of error accumulation by using INS or visual system alone,but also ensure the continuous and usable attitude information in the case of interrupted GPS signals.With the research background of NSFC project “The Research of High Precision Multi-antennas BD/GPS Receiver Modelling and Key Technologies”,the in-depth study of the theoretical error and accuracy analysis of GPS single-antenna attitude determination model was conducted.To improve the accuracy and reliability of the carrier attitude determination,a new GPS/INS/vision attitude determination algorithm is proposed.The main work of this thesis is presented as follows:1.The basic principle of GPS single-antenna attitude determination is studied,and the advantages and disadvantages by using pseudo-attitude to describe the carrier attitude are analyzed.Moreover,the measurement errors,offset errors,and side shift errors are described.Compared with the GPS multi-antenna attitude determination,the kinematic vehicle-borne experiment with GPS single-antenna verified the influence of measurement errors and offset errors.Afterwards,a corresponding correction model was employed.2.For extraction and matching of the feature points in the visual system,we analyze two commonly used algorithms,which are SURF(Speeded Up Robust Features)and ORB(Oriented FAST and Rotated BRIEF).After violent matching,an ORB and matching algorithm on the basis of CUDA(Compute Unified Device Architecture)is proposed,which can enhance the operating efficiency through parallelization.Taking the desktop platform as an example,in the case of 720 P,1080P,and 2K resolutions,the speed at the feature point extraction by using the proposed algorithm is 7.6 times,14.3 times,and 14 times faster;while the speed at matching is 24 times,19.96 times,20.32 times faster.The improvement in feature point extraction and matching is obvious.3.Two methods for attitude determination by using visual system,which are polar constraints based on 2D images and Interceptive C losest Points(ICP)based on 3D images are studied.The former one solves the essential matrix by establishing polar constraints;while the latter one established ICP matching and implement g2o(General Graph Optimization).At last,the Singular Va lue Decomposition(SVD)is applied to acquire the rotation matrix and attitude angles.The experimental results demonstrated that both methods could reflect the change of carrier attitude well.However,due to the errors in the calculation of depth map by using the latter method,better results could be achieved by using the former one.4.The combinations of GPS single-antenna system and INS,GPS single-antenna system and visual system,are introduced respectively to improve the accuracy and reliability of attitude determination.The former combination is a loosely coupled mode that uses the position/velocity differences as a state vector,while the latter one is also a loosely coupled mode that uses the attitude angle as a state vector.Compared to the single system,the kinematic vehicle-borne experiment results showed that the accuracy of attitude determination is improved by using the combined system.Compared with the single INS,the RMS(Root Mean Square)of attitude determination by using combined GPS/INS is reduced by 53.9%;Compared with the single visual system,the RMS of attitude determination by using combined GPS/vision is reduced by 63.6%.This proves the feasibility and superiority of the combination system in attitude determination. |
PDF Full Text Request