| The kind of missions,the complexity of flight environment gradually increasing.Due to the limitations of a single UAV's payload weight and endurance,UAV systems are evolving toward swarming,low-cost,and micro-miniaturization.A single sensor is unable to meet decimeter-level positioning for the UAV swarm requirements.Multi-source fusion positioning technology can improve positioning accuracy by organically coupling various sensors,becoming one of the key directions of UAV swarm positioning.Automatic fast field calibration of UAV systems,collaborative positioning optimization,and rapid fusion positioning for swarm are studied in this paper.The main work and innovation points are as follows:1.Aiming at the automatic fast field calibration of UAV systems,we proposed a zero-speed interval detection model based on differential modulus and a gyroscope field calibration method based on beetle antennae search and particle swarm optimization to enhance the precision of interval detection boundaries and achieve automatic and fast correction for lowcost IMU.The IMU test shows that the calibration time of the proposed method does not exceed 5 minutes,while the calibration time of the existing method based on iterative optimization is about 8 minutes.The calibration parameters are close to those provided by the factory,which can improve the quality of observation information for multi-source fusion positioning.2.To solve the problem that UAV swarm collaborative positioning optimization in urban area,an inverse covariance intersection(ICI)positioning optimization method based on collaborative observation is proposed.The tightly upper bound of the fused positioning estimation error covariance matrix is deduced by the principle of ICI.Based on the minimum variance optimal estimation criterion,the positioning optimization based on nonlinear collaborative observations such as distance is realized,and the positioning accuracy and robustness of the UAV swarm is improved.Simulation results show that compared with the split covariance cross filter(SCIF)and the decorrelation minimum variance method(DMV),the proposed method improves the 95%probability positioning accuracy by 8%and 17.4%,respectively,for the UAV positioning optimization based on relative position and distance measurements.For UAVs with only relative distance measurement,compared with DMV,the proposed method improves the 95%probability positioning accuracy of 12.8%,and the SCIF method cannot optimize the positioning of UAVs.3.In view of the high computational complexity of fusion positioning for UAV swarm,which leads to the difficulty of rapid positioning,a multisource fusion positioning model for low-cost UAV swarm is proposed based on Bayesian networks to high-precision positioning.Aiming at the problems of the calculation complexity of the Bayesian network-based fusion positioning method increasing over time,the communication resource overhead of distributed computing,etc.,a positioning state update method based on distributed matrix decomposition is proposed to reduce communication consumption and computational complexity.A dynamic time window fusion location method based on the kind of condition number rate is proposed to optimize the computational complexity of fusion location and realize fast fusion location.Simulation results show that compared with the fixed time window and AprilSAM method,the proposed method decreases the computation time by 36.1%and 21.5%respectively,and the RMS positioning accuracy is similar.4.We designed a multi-source fusion positioning experiment,based on embedded hardware and a robot operating system to verify the proposed method.The UAV swarm positioning test in the outdoors showed that the RMS fusion positioning accuracy of UAV swarm based on BDS/IMU/magnetometer/UWB/barometer/rangefinder is 0.35m,which can meet the requirements of UAV swarm applications. |
PDF Full Text Request