Autonomous Fault-tolerant Navigation Method Based On Dynamics Model For MUAVs

Multirotor unmanned aerial vehicles(MUAVs)have the advantages of small size,simple structure,easy to control and vertical take-off and landing(VTOL).In recent years,MUAVs have been widely used in many fields,such as pesticide spraying,public safety,environmental monitoring,aerial photography,and so on.Navigation,as the key technology of MUAVs,ensures that the MUAVs can complete tasks and cope with complex environments.However,due to the low accuracy of micro-electromechanical systems(MEMS)inertial measurement units(IMUs)used in the navigation system of the MUAVs,the error of the inertial navigation system(INS)will diverge quickly.Therefore,the navigation information provided by the Global Navigation Satellite System(GNSS)is essential.With the popularity and wide application of MUAV,it will be used in the circumstance of Global Position System(GPS)denied,such as indoor environments,urban blocks with dense high-rise buildings,canyons,jungles,etc.Consequently,when the GNSS information is shielded or interfered,the MUAV navigation system needs other systems to provide accurate,reliable,and robust navigation information,improving the accuracy and autonomy of the navigation system in the complex environment.Based on the above reasons,this dissertation focuses on the research of a MUAV dynamic model(DM),and fuses the DM with the Simultaneous Localization and Mapping(SLAM)technique to improve the navigation accuracy and performance under the circumstance of GPS-denied.Firstly,the MUAV’s navigation technologies under the circumstance of GPS-denied are analyzed in this dissertation.And the researches of light detection and ranging(Lidar),DM and vision sensor aided navigation technology are discussed.The urgent problems needed to be solved for MUAV navigation under the circumstance of GPS-denied are also analyzed.And some key technical issues are summarized when the DM and Lidar are used to aid the navigation system.Secondly,based on the dynamics analysis of the quadrotor,the dynamic model of the quadrotor is constructed.When the parameters of the DM are identified by the traditional wind tunnel test method,the test is complex and the identified parameters are different with the parameters of the actual flight.So an online parameter identification method based on Kalman filter(KF)whose observability and degree of observability are also analyzed,is proposed.The feasibility of the method is proved theoretically.Finally,the actual flight test is used to verify the validity of the method.Using the parameter identification method,the dynamic model can accurately reflect the dynamic characteristics of the quadrotor under the real flight condition,which lays the foundation for the subsequent optimization of DM and the research on DM aided navigation.Thirdly,aiming at the problem that the traditional dynamic drag model can’t accurately reflect the true dynamic characteristics of the quadrotor when it rotates,this dissertation proposes an optimized dynamic drag model with the angular acceleration as a parameter.The model is built based on the analysis of its error characteristics.And its correctness is verified by experiments on a quadrotor.A DM aided navigation method based on KF is also proposed,which is tested on different circumstances.The results of the outdoor flight show that the position estimation accuracy of the improved DM aided navigation compared to the MEMS INS has a significant improvement,with GPS as a reference.And the improved DM aided navigation method also has good localization accuracy in indoor environments without GPS,which provides the theoretical basis for the following research on DM aided autonomous fault-tolerant navigation under the circumstance of GPS-denied.Then,in order to expand the application range of the DM and deal with the problem that two-dimensional Lidar can’t cope with the feature sparse or abrupt environment,an inertial/Lidar autonomous fault-tolerant navigation method based on DM is proposed.It can meet the requirements of reliable localization under the circumstance of GPS-denied.This method uses Federated Kalman Filter(FKF)to fuse the navigation information,and fault diagnosis is based on the residual?~2 test.A simulation platform is built on the Gazebo.And the simulation experiment results show that the method can effectively improve the navigation accuracy when the Lidar SLAM has faults or is not available,and has good reliability and robustness.Finally,a quadrotor platform is built based on DJI M100,and the actual flight experiments are carried out to verify the method presented in this dissertation.The inertial/Lidar autonomous fault-tolerant navigation method based on DM is validated in the feature sparse environment and the abrupt environment.The experimental results show that the proposed autonomous fault-tolerant navigation system still has high accuracy when there are faults on the inertial/Lidar sub-filter,and is reliable.It proves that the method proposed in this paper is correct,effective and has good engineering application value.