GNSS/UWB/VIO Integrated Positioning Theory And Methodology

With the rise of emerging industries,such as the Internet of Things and autonomous driving,modern location-based services are gradually shifting from typical outdoor surveying and mapping to integrated PNT(Positioning,Navigation and Timing)services available in both indoor and outdoor environments.As the commonly used technology in navigation and positioning,Global Navigation Satellite Systems(GNSS)are subject to many challenges arising from the upgrading demand for location-based services.In complex urban environments,the GNSS signal reception can be highly affected by many factors,such as heavy obstruction of buildings or trees and multipath effects.These environmental effects lead to the accuracy,reliability,and continuity of GNSS positioning cannot satisfy users with high precision positioning demands.In addition,compared to traditional high-precision geodetic GNSS receivers,small and low-cost GNSS receivers are more widely applied in urban positioning.However,there are some drawbacks of low-cost receivers challenging the high-precision urban positioning,such as poor quality of observations and more frequent signal loss-of-lock and interruptions.Moreover,in GNSS-denied environments,additional information sources need to be introduced.Ultra-Wide Band(UWB)technology provides millimeter-level precision range measurement,which is appropriate for indoor high-precision positioning.Yet,the performance of existing UWB positioning model is not good enough due to the systematic errors that have not been properly addressed.However,the positioning precision and availability of both GNSS and UWB positioning systems are highly affected by significant signal masking and interruptions.The strategy for introducing sensors with different positioning mechanisms(e.g.,IMU and Cameras)in GNSS/UWB systems is effective to complement or enhance positioning results in outdoor/indoor environments.However,the fusion of GNSS/UWB/IMU/Vision for positioning in complex outdoor and indoor environments is during the start-up phase.There are no sophisticated theoretical fusion algorithms and testing platforms yet.Therefore,it is worth further study on leveraging the sensors with different positioning mechanisms to complement,augment and enhance GNSS positioning for achieving highly available precise indoor and outdoor seamless positioning.For addressing above mentioned research gaps,based on GNSS data processing approaches,this dissertation proposes navigation and positioning methodologies fusing GNSS/UWB/IMU/Vision to improve the precision and reliability of seamless navigation and positioning services in complex indoor and outdoor environments.The main research work and contribution are as follows:(1)Cycle slips detection and repair are important approaches to avoiding reinitializations of integer ambiguities in carrier phase observations.This is crucial to improving the continuity and availability of high-precision GNSS positioning.Currently,the single-frequency and low-cost GNSS receivers are widely used in many urban positioning and navigation applications.However,owing to the incapability of forming between-frequency combination and the relatively poor quality,cycle slip estimation of single-frequency low-cost receivers is a significant challenge.The traditional measurement-based polynomial fitting approaches require historical normal observations without cycle slips from the individual satellite,which is highly vulnerable to failure when cycle slips frequently happen for some satellites.The typical geometry-based(GB)single-frequency cycle slip estimation approaches based on the theory of outlier detection have weak mathematic model strength and low success rate for cycle slip repair.Therefore,a GNSS single-frequency cycle slip estimation method with Positional Polynomial Fitting(PPF)constraint is proposed to address the issues of effective detection cycle clips in complex urban environments.With the assumption that the position of the rover station is subject to a polynomial,PPF estimates cycle slip unknowns using a time-differenced Geometry-based(GB)model with a predicted position constraint to improve the mathematical model strength.The results reveal that,PPF outperforms the traditional methods in cycle slip detection when simultaneous and continuous cycle slips and data interruptions occur.The proposed PPF model is reliable and efficient in real-time kinematic applications with low-cost single-frequency GNSS receivers.(2)A highly reliable single-frequency RTK algorithm,Single-frequency Inexpensive Navigation And Positioning(SINAP),is proposed to address some drawbacks affecting the positioning precision of low-cost GNSS receivers(e.g.,low continuity and low signal-to-noise ratio)in real-time kinematic positioning.In dynamic positioning under complex environments,the accuracy of the float ambiguity unknowns is varying for different satellites due to the frequent signal discontinuities and multipath effects.Considering that float ambiguity unknows are more reliable for continuously observed satellites and less reliable for low-elevation satellites that suffer from relatively large atmospheric errors,a Partial Ambiguity Resolution(PAR)method is proposed.The PAR method selects to fix a subset of the whole ambiguity parameters based on their accumulated time and satellite elevations to improve success rate for Ambiguity Resolution(AR).The performance of SINAP is evaluated using two u-blox M8 T GNSS receivers in complex urban environments with both static and kinematic tests.The experimental results show that the proposed SINAP outperforms the benchmark algorithms(RTKLib and u-blox RTK service)with a higher AR fixing rate and shorter Time-To-First-Fix(TTFF),enables high availability of precise RTK positioning in complex urban environments.(3)Similar to satellite antennas,the offsets between the antenna reference point and the real incidence point defined as the Phase Center Offset(PCO)exist in UWB antennas.However,the PCO is not carefully eliminated from the range observations,which reduces the positioning accuracy.Thus,a Phase Center Offset(PCO)modeling and compensating method for UWB antennas is proposed,and a low-order polynomial function is employed to model the UWB PCO effects.The efficiency of the proposed method is evaluated by a collection of antenna-rotating and tag-moving experiments.The results show that a 2-order polynomial function can compensate for PCO effects by70% in the range observations,efficiently improve the accuracy of time latency determination and positioning.(4)One of the prerequisites of high-precision UWB positioning is to precisely determinate the time latency among UWB anchors.The typical time latency determination methods estimate the time latency among anchors according to the observations collected from a ground reference point with known coordinate.However,the spatially correlated unmodeled errors remaining in the range observations at the calibration point are not carefully considered,which reduces the accuracy of time latency determination and positioning in the whole testing site.Therefore,a Multi-point Time Latency Determination(MTLD)method is proposed,in which multiple calibration points uniformly distributed in the testing site are jointly used in time latency estimation.MTLD can efficiently reduce the impact of spatially correlated errors at a specific point and guarantees higher positioning accuracy overall in the entire area.Based on the experimental results,MTLD improves latency estimation accuracy by 10% and horizontal positioning accuracy by 40%,compared to typical time latency determination methods.(5)In heavily obscured environments,due to the impact of multipath effects and frequent signal interruption,the reliability and availability of GNSS positioning are significantly reduced.The typical strategy for Inertial Navigation System(INS)and GNSS integration cannot guarantee high-precision positioning during long-term GNSS signal outages since INS drifts rapidly along with time.To address this issue,the VisualInertial-Odometry(VIO)navigation enhancement algorithm is proposed.The mathematic model of stereo VIO and the loosely coupled integration of VIO and GNSS/UWB in Earth-Centered Earth-Fixed Frame are derived.The drift error of VIO positioning in case of GNSS signal outage are evaluated via simulated data with different levels of IMU sensors,different accumulated time and accumulated distance.The experimental results show that the drift error of VIO is significantly reduced compared with INS when the long-term GNSS signal outage occurs.Even with the low-cost consumer-grade INS,VIO can still maintain position drift within 1.7 m when GNSS signals are interrupted for 60 s,which effectively improves the availability of position during long-term GNSS signal interruption.(6)In complex environments with long-term GNSS signal outages,the cycle slips are difficult to be successfully detected due to reducing fitting accuracy of both the observations and the position.To address this issue,a Geometry-based VIO-aided(GVIO)GNSS cycle slip estimation method is proposed.This proposed method introduces the constraints of VIO predicted position into the typical GB models,to improve the success rate of cycle slip detection and repair in case of GNSS data interruption.According to the experimental results,for more than 40 s data interruptions,both INS-aided and GNSS-only cycle slip detection methods fail.In contrast,the VIO-aided detection method still achieves a 40% ～50% success rate of cycle slip and data gap repair.(7)Proposes a tightly-couple integration algorithm of GNSS/UWB/INS/Vision observations to provide high-precision and highly available outdoor and indoor seamless positioning.Also,a multi-sensor integrated data collection platform and GNSS/UWB/INS/Visual-Odometry integrated navigation and positioning software system are developed.The seamless indoor-outdoor positioning system is evaluated in the complex environment of the campus.The experiment results show that the tight combination of four sensors can achieve seamless indoor and outdoor positioning with decimeter-level overall accuracy in complex environments.Compared with GNSS/UWB positioning and GNSS/UWB+INS positioning,the fusion of GNSS/UWB/VIO shows a significant improvement in the accuracy and continuity of positioning results,especially in the situation of heavy signal blockage and interruptions.