Key Technology For Real-Time Vehicle Precision Positioning In Complex Urban Scenes

With the rapid development of the intelligent driving industry,the demand of precise positioning for land vehicle is higher.The GNSS and SINS integrated navigation system has the advantages of two types of sensors,and has been utilized in weapon guidance,space exploration,geodesy,etc.GNSS / SINS integrated navigation system can meet most engineering needs in open scenarios,but the performance of navigation ststem still faces many problems in the complex urban environment.It is becoming the key factor for the widespread promotion.Based on this research background,the objective of this paper is key technologies of GNSS / SINS integrated navigation for land vehicle in urban complex environments.The main research work is as follows:1)The characteristics and data quality of dynamic GNSS observations in urban environments is analyzed.The detail analysis is executed from data visibility,carrierto-noise,the accuracy of pseudorange residuals,LLI identifiers,etc.Then the GNSS observation random model is refined using the carrier-to-noise,and an improved GNSS data preprocessing strategy is proposed.Based on the receiver LLI identifier and the consistency among phase,pseudorange,and Doppler observations,the ability of pseudorange gross error and phase cycle slip detection is improved.The results of real vehicle test data show that the RMS of the plane position can reach 0.13 and 0.14 m.The accuracy is improved 53.7% and 58.8% compared to traditional methods of considering the elevation angle only and detecting cycle slips using the combination of MW and GF.2)Ambiguity-fixed resolution is a prerequisite for obstaining precise positioning results.Based on the GNSS measurement analysis and data processing experience,a partial ambiguity fixing method with multi-quality-control is proposed.In this paper,the ambiguity-subset selection strategy is optimized,and the multi-quality-control is divided into two steps: for one ambiguity before integer search and for the fixing results after integer search.Two sets of real vehicle test data are used to verify the proposed method.The correct ambiguity-fixed rates is 75.28% and 69.0%,respectively.And this value is 55.86% and 46.98% when the full ambiguity resolution is adopted.Using the improved method,the positioning error RMS of ENU is 6.99,9.89 and 3.94 cm.It is improved by 33%,2% and 36% compared with the FAR.The results show that this method can effectively improve the ambiguity fixing performancing in dynamic applications.3)In this paper an ambiguity fixing method with ionosphere predication constrained is proposed.Based on this method,historical results are used to calculate ionospheric delays.Then a linear model is used to model and predict it when taking into account the spatio-temporal characteristics of the ionosphere.Then the ionospheric predication would be used to constrain the relative positioning model for obstaining the ambiuiguity-fixed resolution fastly.A set of 92 km vehicle data is used to verify the proposed method.The results show that ionospheric prediacation can assist in fixing the ambiguity.Using the proposed method,the ambiguity fixing rate can reach88.92%,and the positioning RMS of ENU is 0.02,0.03 and 0.08 m.While the fixing rate of the ionosphere free combination is only 21.60% and the positioning RMS is 0.75,0.71 and 1.90 m.Obviously,the positioning accuracy of the new method is improved by 97.3%,95.8% and 95.8%.4)In this paper,a loosely-dynamic-constraint model is proposed.Considering the characteristics of vehicle,raw IMU data is used to identify the vehicle motion state.Then based on the detection results,the appropriate constraints could be selected.At last,a tightly coupled scheme of GNSS/SINS/vehicle motion constraint is built.The results show that the vehicle motion detection method proposed in this paper can effectively distinguish the vehicle motion status.The results of the simulated tunnel scene show that this method can suppress the drift of inertial navigation when the GNSS signals are outage.And the percentage of plane position error drift and driving distance is recursively derived from pure inertial navigation 5.69% decreased to0.56%.In addition,a section of severe occlusion environment data of a vehicle driving under an overpass is seclect.The results show that the GNSS/SINS/motionconstrained fusion scheme can provide a positioning result of 1.82 m in plane.Compared with the traditional GNSS/SINS tight coupled method,the positioning accuracy in the plane is improved by 81.6%.5)A set of GNSS/SINS integration algorithm software platform is realized,and a multisensor hardware acquisition platform is built.Two sets of dynamic data collected in complex urban environment are used to verify the algorithm.The results show that the algorithm is operating normally.The solutions only using GNSS can ensure more than 90% avalibality.The GNSS/SINS integrated system can provide better positioning results,and the accuracy is better than 0.5m in the plane direction in urban scenes.More than 80% of the plane positionings are better than 0.5m and the maximum positioning error is less than 2m.Even in severe occlusion environment which is blocked by trees and buildings,the plane accuracy is still within 2m.And the accuracy of speed and More than 66% of the plane positionings are better than 1m.The accuracy of velocity and attitude are basically the same as that of commercial software.