Research On Integrated Navigation System Based On Inertial/Bionic Vision

In order to meet the demand for navigation systems during long-duration operations of unmanned combat platforms,most unmanned platforms use an inertial navigation system(INS)and a global satellite navigation system(GNSS)to effectively improve Long-term navigation performance.However,GPS is a passive navigation system,which is susceptible to signal jamming or failure caused by human factors or external environmental factors,which affects the overall performance of the integrated navigation system to a certain extent.At the same time,most of the existing inertial-based integrated navigation systems use auxiliary sensors and INS for information fusion,and the entire navigation system needs to be improved in terms of intelligence.In order to study the intelligent navigation method,the researchers turned their attention to the biological world,and carried out the research on the bionic navigation method and system.Based on an in-depth analysis of the animal navigation mechanism,this paper focuses on the data fusion algorithm based on Kalman filtering,the inertial/visual matching combined navigation method based on the location cell navigation model,and the high real-time algorithm implementation method based on GPU accelerated parallel processing.The feasibility of the research results in practical application is verified by experiments.The thesis mainly researches from the following aspects:(1)The imaging bionic polarized light compass algorithm is studied and a hardware system is built.Firstly,the atmospheric scattering model is deeply analyzed and discussed,and the Rayleigh scattering model is studied.The polarization description method based on the Stokes vector is given.And the classic Stokes parameter measurement method is obtained;Secondly,the polarization light heading angle measurement algorithm is introduced,and the solar azimuth angle calculation method,the polarization angle measurement method and theheading angle calculation method are described;Finally,the Nvidia Jetson TX2 is used to build a polarized light measurement system hardware platform.In the whole system,Nvidia Jetson TX2 undertook the system global control work and heading angle calculation.Dynamic experiment verification is performed.The results show that the root mean square error of the three experiments is better than 0.42°,proving the superiority of the imaging-type bionic polarized compass system.(2)A bionic polarized light/inertial combined heading measurement method is proposed.Firstly,the definition and parameter description of common coordinate systems are briefly described.Then,the attitude of the inertial device is estimated by the quaternion method.Finally,the Kalman is proposed for the problem of the accumulation of the inertial device's long endurance error and the artificially given initial heading angle.The filter fusion algorithm makes the inertial device and the polarized light device have complementary advantages.An experimental platform is built to verify the above theory.The navigation accuracy is improved,and the feasibility of the above method is verified.(3)Inertial/visual matching combined navigation method based on location cell navigation model.Firstly,the characteristics of the position cell are briefly analyzed.Secondly,the SLAM_ORB feature extraction is deeply analyzed and discussed from the aspects of concept,features and methods,and the advantages and disadvantages of the SLAM_ORB algorithm are analyzed,which reduces the uncertainties in the shooting process.The noise effect of the image has been reduced and enhanced on the image of the position cell node,and the SLAM_ORB algorithm has been improved to greatly increase the matching accuracy;then the inertia is calculated and the speed and position are obtained.The position cell node is used to correct the position error of the inertial device,which reduces the speed and position divergence of the inertial system;in order to solve the problem of real-time matching of the position cell node,the GPU's advantages in processing images are described,and the CUDA acceleration method is proposed;Secondly,an in-vehicle experimental platform of an inertial/visual combination system is built and the experiments are carried out.The results show that,compared with the pure inertial navigation mode,the method proposedin this paper effectively improves the navigation accuracy and reduces the position measurement error by more than 60%.The inertial/bionic visual integrated navigation system proposed in this paper enriches and expands modern navigation theory,and provides technical support for the unmanned ground combat platform to successfully execute combat missions.