Analysis And Algorithm Design Of Multi-vehicle Cooperative Positioning Technology

Cooperative localization utilizes heterogeneous agent to exchange information within a neighborhood and achieve individual localization in a formation through information fusion.As a common carrier for agent,wheeled vehicle localization uses a centralized structure to process multiple sensor measurements to improve accuracy.However,using sensors to collect information for multi-vehicle system location in the GNSS rejection region has stringent constraints for the type and quantity of information.Vehicle localization based on the Kalman filtering algorithm in an outdoor environment will have the problems of low matching between vehicle motion model and actual model and the measurement values owning multiplicative noise.How to solve the problems of motion model mismatch,measurement error fluctuation,and improve the positioning accuracy of the multi-vehicle system has become an urgent problem for multi-vehicle cooperative positioning.In this paper,we adopt a multi-vehicle cooperative positioning scheme based on wireless sensor network technology and increase the information redundancy to improve the estimation accuracy by using information exchange to address the problem of large single-vehicle positioning errors.In the face of the outdoor real environment,the filtering algorithm is improved accordingly,and the problems that the information collected by the sensor is affected by multiplicative noise and the mismatch between the motion model and the actual model are solved one by one.The performance of the average consistency-based and maximum consistency-based localization algorithms is compared and analyzed by simulation.The dissertation focused on the following research accomplishments.（1）The traditional cubature Kalman filtering algorithm is improved for the measured values contain correlated multiplicative noise.By constructing a multiplicative noise model,the traditional Gaussian filtering algorithm is improved by using the innovation analysis method,and extended to the cubature Kalman filtering algorithm based on the known and unknown noise correlation coefficients to obtain an adaptive cubature Kalman filtering algorithm.（2）An interactive multi-model filtering algorithm is used to replace the single-model filtering algorithm to improve the model matching.By combining it with the adaptive cubature Kalman filtering algorithm,an interactive multi-model adaptive cubature Kalman filtering algorithm is proposed to solve the problem of low localization accuracy of single-model filtering algorithm in the outdoor environment.（3）In the design of the cooperative localization algorithm,the information transfer relationship between heterogeneous vehicles is constructed by using a directed strong connection graph in graph theory,and a distributed filtering algorithm is proposed by combining the idea of a maximum consistency with cubature Kalman filtering to solve the problem of slow convergence of the information consistency process.The interactive multimode adaptive cubature Kalman filtering is used to replace the cubature Kalman filtering algorithm in the distributed filtering algorithm to improve the state estimation accuracy of the distributed filtering algorithm and applies to the Mecanum wheel vehicle for simulation.