GNSS / INS / vision tightly coupled system for long coherent integration of gnss carrier phase tracking

Abstract

The application provides a GNSS / INS / Vision deep combination system for GNSS carrier phase long coherent integration tracking, which comprises an upper-layer multi-source data fusion subsystem, a bottom-layer GNSS baseband signal processing subsystem and a deep combination auxiliary module, the upper-layer multi-source data fusion subsystem generates multi-source dynamic auxiliary Doppler information through the deep combination auxiliary module and inputs the multi-source dynamic auxiliary Doppler information to the bottom-layer GNSS baseband signal processing subsystem, so as to eliminate the dynamic stress generated by the relative motion between the receiver and the satellite in the coherent integration process; and the bottom-layer GNSS baseband signal processing subsystem provides pseudorange, Doppler and carrier phase observations for the upper-layer multi-source data fusion subsystem, and is used for calculating the navigation state of the receiver. The application can improve the continuity and availability of GNSS carrier phase in a complex environment, and can provide continuous centimeter-level positioning in the complex environment, so as to meet the precise positioning and navigation requirements of robots and autonomous vehicles in the complex environment.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of GNSS receivers, and particularly relates to improving the continuity of GNSS carrier phase in complex environments and the availability of high-precision positioning. BACKGROUND

[0002] With the rapid development of automation technology, the navigation demand of robots and autonomous vehicles in complex environments is increasing day by day. Global Navigation Satellite System (GNSS) can provide users with all-weather, high-precision Position Velocity and Time (PVT) information, and is an indispensable sensor in navigation and positioning solutions. However, in complex environments such as cities and shaded roads, due to problems such as signal attenuation, reflection and obstruction, the precision and continuity of GNSS carrier phase are severely degraded, greatly limiting the availability of high-precision GNSS positioning.

[0003] Typical high-precision GNSS positioning technologies include Real-time Kinematic (RTK), Precise Point Positioning (PPP) and Precise Point Positioning-Real-Time Kinematic (PPP-RTK). The above algorithms all require continuous and accurate carrier phase observations. Carrier phase observations are generated by a Phase-locked loop (PLL) in a GNSS receiver continuously tracking GNSS carrier signal components. Traditional PLLs are extremely susceptible to small signal disturbances in complex environments, resulting in frequent or even interrupted GNSS carrier phase observations, which severely deteriorates positioning accuracy and continuity. Loosely coupled and tightly coupled technologies respectively perform data fusion at the level of positioning results and at the level of observations (pseudo-range, Doppler, carrier phase), which can improve the availability and continuity of positioning to some extent, but both cannot improve the continuity of GNSS carrier phase observations and cannot fully exploit the potential of GNSS.

[0004] In complex environments, signals suffer from blockage, non-line-of-sight (NLOS) and multipath effects, and the traditional GNSS receiver carrier phase tracking robustness is poor. Humphreys studied the carrier phase vector tracking method, but this method is extremely dependent on the top layer of precise positioning, and will fail when only precise positioning is not available, and it is easy to fail in complex environments. Li et al. used a multi-channel co-operative tracking loop (Co-Op) to estimate the loop clock drift, which can reduce the influence of oscillator instability in the tracking process. O Driscoll et al. proposed a long coherent integration tracking method, which can to some extent alleviate the multipath effect and improve the signal energy. These methods can improve the carrier phase tracking performance to some extent, among which the long coherent integration structure can be compatible with the traditional receiver and has the best performance. However, the trade-off between dynamic stress and thermal noise suppression in GNSS tracking methods without external information assistance has not been completely solved, and the coherent integration time cannot be effectively extended. The deep combination technology of introducing external sensor assisted baseband signal processing can effectively reduce the influence of dynamic stress on the loop. Soloviev et al. proposed a GNSS / INS deep combination solution, which can achieve sub-meter positioning accuracy in dense forest areas where traditional receiver technology cannot track GNSS signals. Ren et al. used maximum likelihood method to estimate navigation bits, and realized long coherent integration through vector tracking structure deep combination assistance. Bochkati et al. proposed a synthetic aperture method, which compresses the bandwidth for synthetic antenna aperture processing (SAP), improves the directivity of the signal and reduces the multipath interference, but the error of INS accumulated over time will seriously reduce the system performance. The introduction of vision and other technologies can effectively improve the INS recursion ability when GNSS fails, and significantly improve the dynamic stress assistance effect. Zuo et al. explored the GNSS / IMU / Vision multi-source deep combination method, but only provided preliminary simulation results. Multi-source deep combination still has broad research prospects, and there is currently no work on multi-source deep combination research for carrier phase tracking.

[0005] The method of multi-sensor fusion assisted baseband carrier phase tracking can fully suppress the influence of dynamic on the tracking loop, solve the contradiction between dynamic stress and thermal noise suppression from the root, thereby prolonging the coherent integration time of the tracking loop, suppressing thermal noise, and having important significance for stable and continuous tracking of carrier phase in complex environments and reliable precise positioning. SUMMARY

[0006] The purpose of the present application is to provide a GNSS / INS / Vision deep combination system for GNSS carrier phase long coherent integration tracking, which can improve the availability of GNSS carrier phase in complex environments and meet the navigation needs of robots and autonomous vehicles in complex environments.

[0007] The technical scheme of the present application provides a GNSS / INS / Vision deep combination system for GNSS carrier phase long coherent integration tracking, comprising an upper-layer multi-source data fusion subsystem, a bottom-layer GNSS baseband signal processing subsystem and a deep combination auxiliary module,

[0008] The upper-layer multi-source data fusion subsystem generates multi-source dynamic auxiliary Doppler information through the deep combination auxiliary module and inputs the information to the bottom-layer GNSS baseband signal processing subsystem to eliminate dynamic stress generated by the relative motion of the receiver and the satellite in the coherent integration process.

[0009] The bottom-layer GNSS baseband signal processing subsystem provides pseudorange, Doppler and carrier phase observations for the upper-layer multi-source data fusion subsystem, which are used to calculate the navigation state of the receiver.

[0010] Moreover, the upper-layer multi-source data fusion subsystem comprises a GNSS precise positioning module, an inertial navigation module, a visual navigation module and a multi-source data fusion module.

[0011] Moreover, the multi-source data fusion module uses GNSS positioning results from the GNSS precise positioning module, mechanical programming results from the inertial navigation module and feature point triangulation results from the visual navigation module as inputs, uses a multi-state constraint Kalman filter to fuse multi-source data and outputs high-precision fusion positioning results centered on the inertial navigation.

[0012] Moreover, the bottom-layer GNSS baseband signal processing subsystem comprises a signal preprocessing module, a long coherent integration control module and a carrier phase phase-detecting control module.

[0013] Moreover, in the signal preprocessing module, after the intermediate frequency signal is stripped of the carrier and the pseudo-random code through a mixer and a correlator, I and Q channel correlation results are generated and input to the long coherent integration control module for coherent integration of the navigation bit length.

[0014] Moreover, the long coherent integration control module reads navigation bits through an external bit auxiliary interface and performs long-time coherent integration of multiple navigation bit lengths based on the navigation bit transmission time after frame synchronization is valid.

[0015] Moreover, the long coherent integration control module adopts a multi-stage convergence mode, gradually extends the coherent integration time and compresses the loop bandwidth to achieve the target coherent integration time.

[0016] Moreover, the carrier phase phase-detecting control module supports two-quadrant and four-quadrant phase-detecting mode switching, adopts the two-quadrant phase-detecting mode in the initial stage and switches to the four-quadrant phase-detecting mode using external navigation bit assistance when the loop enters a stable tracking state.

[0017] Moreover, the deep combination auxiliary module generates multi-source dynamic auxiliary Doppler information based on receiver dynamics and satellite dynamics, the receiver dynamics is obtained by fusing positioning results and compensating for antenna boom arms, and the satellite dynamics is obtained by calculating ephemeris by a GNSS precise positioning module; the multi-source dynamic auxiliary Doppler information is input to a bottom-layer GNSS baseband signal processing subsystem after being extrapolated by a uniform acceleration model, so as to eliminate dynamic stress.

[0018] In another aspect, the application also provides a positioning method implemented by a GNSS / INS / Vision deep combination system facing GNSS carrier phase long coherent integration tracking, for precise positioning and navigation of robots and autonomous vehicles in complex environments.

[0019] The application can improve the continuity and availability of GNSS carrier phase in complex environments, and provide continuous centimeter-level positioning in complex environments, thereby meeting the precise positioning and navigation requirements of robots and autonomous vehicles in complex environments. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings of the embodiments will be briefly introduced below.

[0021] Figure 1 The figure is a whole block diagram of the multi-source deep combination system structure of the embodiments of the application.

[0022] Figure 2 The figure is a principle block diagram of MSCKF fusion positioning calculation of the embodiments of the application.

[0023] Figure 3 The figure is a deep combination mode switching principle block diagram of the embodiments of the application.

[0024] Figure 4 The figure is a comparison chart of loop tracking Doppler and loop estimation Doppler of BD13 satellite in the embodiments.

[0025] Figure 5 The figure is a comparison chart of multi-source auxiliary Doppler of BD13 satellite before and after extrapolation in the embodiments.

[0026] Figure 6 The figure is a baseband loop tracking principle diagram of the embodiments of the application.

[0027] Figure 7 The figure is a comparison chart of two-quadrant and four-quadrant traction ranges of the embodiments of the application.

[0028] Figure 8 The figure is a comparison chart of two-quadrant and four-quadrant discrimination results of BD13 satellite in the embodiments.

[0029] Figure 9Fig. 20 and Fig. 30 respectively represent the discriminator output diagrams corresponding to the 20ms and 300ms coherent integration time lengths for the BD13 satellite in the embodiment.

[0030] Figure 10 Fig. 20 and Fig. 30 respectively represent the carrier phase error diagrams corresponding to the 20ms and 300ms coherent integration time lengths for the BD13 satellite in the embodiment.

[0031] Figure 11 Fig. 20 and Fig. 30 respectively represent the horizontal positioning error diagrams corresponding to the 20ms and 300ms integration time lengths for the traditional commercial receiver and the multi-source deep combination system in the embodiment. DETAILED DESCRIPTION

[0032] In order to more clearly illustrate the technical solutions in the present application and / or the prior art, the specific embodiments of the present application will be described below with reference to the drawings. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor, and other embodiments can also be obtained.

[0033] Embodiment 1

[0034] The following will take the carrier phase tracking of the long-time coherent integration of the multi-source deep combination system in the dynamic scenario of the GNSS receiver as an example, and the present application will be described in detail in combination with the drawings. The drawings and the specific embodiments are exemplary for the present application, and are not limiting to the present application.

[0035] The overall structure of the multi-source deep combination system provided by the embodiment of the present application is shown in Fig. 1, which comprises an upper-layer multi-source data fusion subsystem, a bottom-layer GNSS baseband signal processing subsystem and a deep combination auxiliary module. Figure 1

[0036] The upper-layer multi-source data fusion subsystem comprises a GNSS precise positioning module, an inertial navigation module, a visual navigation module and a multi-source data fusion module.

[0037] The GNSS precise positioning module inputs the observation information of each satellite output by the GNSS baseband signal processing subsystem, and uses a centimeter-level positioning algorithm. It is proposed in the present application that the high-precision positioning algorithms such as RTK, PPP and PPP-RTK can be used for positioning of the module, and the GNSS high-precision positioning result is output for subsequent combination.

[0038] The GNSS precise positioning module inputs the observation information of each satellite output by the GNSS baseband signal processing subsystem, and uses a centimeter-level positioning algorithm. It is proposed in the present application that the high-precision positioning algorithms such as RTK, PPP and PPP-RTK can be used for positioning of the module, and the GNSS high-precision positioning result is output for subsequent combination.

[0039] ​The inertial navigation module performs inertial navigation mechanical orchestration based on raw data from the IMU's accelerometers and gyroscopes. This module propagates navigation state and outputs the vehicle's attitude, velocity, and position information. The visual navigation module uses camera data for feature point extraction, tracking, and triangulation, which is then fused with the inertial navigation mechanical orchestration.

[0040] The multi-source data fusion module uses GNSS positioning results from the GNSS precise positioning module, mechanical orchestration results from the inertial navigation module, and feature point triangulation results from the visual navigation module as inputs. It uses a multi-state constrained Kalman filter (MSCKF) to fuse the multi-source data, thereby outputting a high-precision fused positioning result centered on the inertial navigation system, as required by this invention. Specifically, the MSCKF fused positioning solution propagates the system error state through the inertial navigation mechanical orchestration. When a camera keyframe arrives, constraints are established for multiple cameras observing the same visual feature point, a visual observation equation is constructed, and the Kalman filter is updated. When the GNSS positioning result arrives, a GNSS position observation equation is constructed, and the Kalman filter is updated. The error state updated by the Kalman filter is fed back to the inertial navigation module to update the IMU error.

[0041] The underlying GNSS baseband signal processing subsystem includes a signal preprocessing module, a long coherence integration control module, and a carrier phase detection control module.

[0042] in,

[0043] In the signal preprocessing module, the intermediate frequency signal is fed into a mixer and mixed with the carrier signal generated by the local voltage-controlled oscillator. The mixed signal is then fed into a correlator for 1ms correlation to obtain the correlation results for the I and Q paths, which are then input into the long coherent integration control module. After bit synchronization is effective, the I and Q path correlation results are used to perform coherent integration for one navigation bit length, starting from the navigation bit edge.

[0044] The present invention further proposes:

[0045] The long-coherence integration control module reads the navigation bits through an external bit auxiliary interface. After frame synchronization is effective, it performs long-time coherence integration on the current bit based on the navigation bit transmission time, which is the length of multiple navigation bits.

[0046] Through multi-stage convergence, the coherent integration time is gradually extended, and the loop bandwidth is compressed to achieve the target coherent integration time. The initial coherent integration time is 1 navigation bit length, and the target coherent integration time is N navigation bits length. Multiple loop convergence stages are set between the initial and target coherent integration times. A phase-locked indicator is calculated for each loop update. Set PLI threshold. When 20 times of continuous detection the loop is judged to converge, and the next stage is switched to until the target coherent integration time is reached.

[0047] The carrier phase discriminator control module uses the same external bit auxiliary interface as the long coherent integration control module, and under the assistance of external bits, the two- and four-quadrant phase discrimination modes can be switched. The initial loop uses the two-quadrant phase discrimination mode, and after successfully obtaining a fixed solution and continuously positioning for 6 to 10 epochs, it is considered that the loop enters a stable tracking state, and the external navigation bit is used to switch to the four-quadrant phase discrimination mode.

[0048] The upper-layer multi-source data fusion subsystem provides high-precision receiver and satellite positions, which are sent to the deep combination auxiliary module to generate multi-source dynamic auxiliary Doppler information, which is sent to the bottom-layer GNSS baseband signal processing subsystem to eliminate the dynamic stress caused by the relative motion of the receiver and the satellite during the coherent integration process. Specifically, the inputs of the deep combination auxiliary module include receiver dynamics (position, velocity, acceleration) and satellite dynamics (position, velocity), which are provided by the upper-layer multi-source data fusion subsystem. The high-precision fusion positioning result is obtained after antenna rod arm compensation, and the satellite dynamics is obtained by solving the broadcast ephemeris or precise ephemeris in the GNSS positioning module. The Doppler frequency shift caused by the relative motion of the receiver and the satellite is calculated through the dynamic information of the receiver and the satellite, and the multi-source auxiliary Doppler is generated. The present application proposes that the uniform acceleration model is used to linearly extrapolate the multi-source auxiliary Doppler information, which is sent to the bottom-layer GNSS signal processing subsystem after extrapolation to 1000Hz, to eliminate the dynamic stress caused by the relative motion of the receiver and the satellite during the coherent integration process.

[0049] Preferably, the loop initially works in a 20ms integration mode, the comparison threshold is set to 5Hz-10Hz, the clock drift output by the positioning solution module is combined with the multi-source auxiliary Doppler to estimate the loop Doppler, and the difference between the two is compared with the loop output Doppler. When the difference is less than the comparison threshold, it is considered that the multi-source auxiliary Doppler information is effective, and the deep combination mode using multi-source auxiliary Doppler for compensation is switched to.

[0050] The bottom-layer GNSS signal processing subsystem provides pseudorange, Doppler, and carrier phase observations for the upper-layer high-precision GNSS positioning module to calculate the receiver navigation state, wherein the observation extraction frequency is adjustable from 1 to 50Hz.

[0051] Referring to Figure 1The overall working process of the system is that: the intermediate frequency signal generates observation values of each satellite channel through the GNSS baseband signal processing subsystem, and the observation values are used for positioning calculation of the GNSS precision module. The visual inertial odometer module performs inertial navigation mechanical arrangement and feature point extraction, tracking and triangulation of camera data. The multi-source data fusion module fuses the inertial navigation mechanical arrangement results, visual feature measurement results and GNSS high-precision positioning results through MSCKF to obtain high-precision multi-source fusion navigation positioning results. In the deep combination auxiliary module, the multi-source fusion results and satellite position information are used to calculate the auxiliary Doppler information of the deep combination, and the Doppler auxiliary information is linearly extrapolated at 1000Hz to assist different loops of the baseband tracking, control the output of the loop NCO and eliminate the influence of dynamic stress.

[0052] Embodiment 2

[0053] On the basis of the system provided in Embodiment 1, the implementation manner of the MSCKF fusion positioning result calculation is as follows:

[0054] A multi-state constraint Kalman filter (MSCKF) is used to perform top-level positioning fusion of GNSS, INS and vision. The principle diagram of the MSCKF fusion positioning calculation is as shown in Figure 2 This part can be implemented by referring to the prior art. The inertial navigation mechanical arrangement performs navigation state propagation, and establishes observation equations of visual update and GNSS position update to perform measurement update. On the basis of the existing MSCKF fusion positioning calculation technology, the present application proposes an overall framework with inertial navigation as the central axis, establishes visual observation models and GNSS position observation models through inertial navigation recursion, and feeds back the state of the MSCKF update to the inertial navigation recursion for updating the IMU error.

[0055] First, the IMU state vector is augmented by the camera pose. After feature point extraction and inertial navigation mechanical arrangement recursion, the data at the whole second time and the camera sampling time are used for state augmentation:

[0056]

[0057] In formula (3):

[0058] The augmented state vector is represented.

[0059] The state of the IMU at the epoch is represented.

[0060] ]… ] represents the first camera pose in the window.

[0061] MSCKF uses multiple frames of feature observations within a window to update the state. When the visual feature tracking is interrupted or the continuous tracking time exceeds the length of the sliding window, the MSCKF is updated by triangulating the features with augmented state variables, constructing the feature point observation equations and projecting into the null space, and then performing the MSCKF update. In the first step of the MSCKF update, the state variables are augmented with the feature point positions in the navigation frame. The predicted pixel coordinates of the th feature point in the frame image can be expressed as:

[0062]

[0063] In equation (4), represents the feature point position in the navigation frame.

[0064] and represent the IMU pose and position, respectively.

[0065] and represent the camera-IMU extrinsic parameters, respectively.

[0066] K and represent the camera intrinsic and distortion parameters, respectively.

[0067] represents the transformation matrix that transforms the feature point in the navigation frame to the camera frame.

[0068] and represent the projection and distortion processing functions in the image processing process, respectively.

[0069] The observation model of the visual feature point can be obtained by perturbation analysis of the IMU state and the feature point position:

[0070]

[0071] In equation (3),

[0072] represents the pixel observation value.

[0073] represents the perturbation term of the IMU state.

[0074] represents the perturbation term of the feature point.

[0075] represents the Jacobian matrix of the IMU state perturbation.

[0076] represents the Jacobian matrix of the feature point position perturbation.

[0077] represents observation noise.

[0078] To reduce the computation in the algorithm, the term in equation (3) is removed using left null space projection, then the measurement model of visual feature points is simplified as:

[0079]

[0080] where, represents the observation matrix after null space projection, represents the observation noise after null space projection.

[0081] When the GNSS positioning result is updated, the position observation information is calculated using the IMU pose at the corresponding time, so as to correct the system position. The relationship between the predicted antenna position and the position calculated by the IMU is as follows:

[0082]

[0083] In equation (5),

[0084] represents the predicted antenna position.

[0085] represents the direction cosine matrix from b system to n system.

[0086] represents the position of the IMU.

[0087] represents the rod arm between the antenna and the center of the IMU.

[0088] Similarly, disturbance analysis is performed to derive the GNSS position observation model:

[0089] In equation (6),

[0090] represents the residual error of the GNSS position observation.

[0091] represents the rod arm between the antenna and the center of the IMU.

[0092] and respectively represent the disturbance terms of the inertial navigation position and attitude.

[0093] is the GNSS observation noise.

[0094] ​The system state can be updated using the visual observation equation and the GNSS position observation equation constructed above, and a high-precision fusion navigation result can be output after updating.

[0095] Embodiment 3

[0096] On the basis of the system provided in Embodiment 1, a loop dynamic stress compensation implementation manner is as follows:

[0097] The application proposes that, in the MSCKF fusion positioning module, the inertial navigation mechanical arrangement is taken as a central axis to provide dynamic auxiliary information for deep combination, when a new key frame or GNSS positioning result arrives, the Kalman filter is updated, and the error of the IMU is corrected, so that continuous high-precision fusion positioning results are output.

[0098] In this embodiment, the high-precision fusion positioning result output by the MSCKF fusion positioning module provides the pose, velocity and acceleration information of the inertial navigation, the dynamic information of the receiver is obtained through antenna arm compensation, the satellite position motion information from the GNSS positioning solution module is combined, multi-source auxiliary Doppler is solved and is assisted to different satellite channels in the baseband, the influence of dynamic stress is eliminated, so that the bandwidth is compressed and the integral time is extended.

[0099] In this embodiment, the deep combination multi-source auxiliary Doppler The Doppler caused by the receiver motion and the Doppler caused by the satellite motion are composed, as follows:

[0100]

[0101] In formula (7):

[0102] represents the motion Doppler generated by the first satellite.

[0103] represents the Doppler generated by the receiver motion.

[0104] Wherein the Doppler of the carrier motion can be calculated by the following formula:

[0105]

[0106] In formula (8):

[0107] is the unit vector in the line-of-sight direction between the carrier and the satellite in the ECEF coordinate system.

[0108] is the motion velocity of the carrier in the ECEF coordinate system.

[0109] is the wavelength of the satellite signal.

[0110] where may be calculated by the following formula:

[0111]

[0112] In formula (9):

[0113] is the position of the first satellite provided by the GNSS positioning module.

[0114] is the high-precision receiver position information provided by the MSCKF.

[0115] Doppler caused by satellite motion is calculated as follows:

[0116]

[0117] In formula (10): represents the velocity of satellite motion.

[0118] In this embodiment, in order to ensure that the provided dynamic assistance information can effectively eliminate the influence of dynamic stress in time, the output rate of the provided assistance information should be high enough. Therefore, the calculated motion Doppler needs to be linearly extrapolated. In this embodiment, a uniform acceleration extrapolation model is adopted for both satellite motion Doppler and receiver motion Doppler.

[0119] For the Doppler caused by carrier motion, the extrapolation method is as follows:

[0120]

[0121] In formula (11):

[0122] and respectively represent the Doppler frequency shift caused by carrier motion at two consecutive update times.

[0123] represents the acceleration of receiver motion.

[0124] represents the update period of receiver position and velocity information.

[0125] represents the update period of the baseband loop.

[0126] In the above formula, the acceleration of receiver motion is obtained by projecting the acceleration of high-precision fusion positioning results: ​

[0127]

[0128] In equation (12):

[0129] and denote the antenna and inertial acceleration in the navigation frame, respectively.

[0130] denotes the direction cosine matrix from the body frame to the navigation frame.

[0131] denotes the inertial angular velocity in the body frame.

[0132] denotes the derivative of the inertial angular velocity.

[0133] denotes the antenna boom.

[0134] The acceleration in the navigation frame is then projected:

[0135]

[0136] In equation (13):

[0137] denotes the acceleration of the receiver motion in the Earth-Centered, Earth-Fixed (ECEF) frame.

[0138] denotes the direction cosine matrix from the navigation frame to the ECEF frame.

[0139] For the Doppler due to satellite motion, the extrapolation method is as follows:

[0140] In equation (14):

[0141] and denote the Doppler shift due to satellite motion between two consecutive updates.

[0142] denotes the difference in satellite Doppler between two update times.

[0143] denotes the positioning period.

[0144] Figure 3The diagram illustrates the principle block diagram of the system's deep combination mode switching. In the GNSS baseband signal processing subsystem, the input intermediate frequency signal is preprocessed for different satellite channels, then loop-filtered before being input to the NCO for control. Multi-source auxiliary information can replace the Doppler information from local loop tracking. NCO is controlled. This invention takes into account the Doppler information tracked by the PLL loop. It includes both dynamic stress and receiver clock drift. Therefore, when using multi-source assisted Doppler information to control the loop, it is necessary to use the receiver clock drift output from the positioning solution module and convert it into Doppler information. Doppler estimates of a loop composed of multiple source-assisted Dopplers Compare it with the Doppler output of the actual loop tracking. In comparison, when the difference between the two is less than the comparison threshold of 5-10Hz, the multi-source auxiliary Doppler information is considered valid, and the system switches to a deep combination mode that uses multi-source auxiliary Doppler for assisted tracking. Figure 4 The comparison between loop estimation Doppler and loop tracking Doppler for strong signals is shown. It can be seen that the two trends are highly consistent. Most of the dynamic stress in the loop can be eliminated by multi-source assisted Doppler. The loop tracks the residual Doppler after assistance.

[0145] In this embodiment, the multi-source auxiliary Doppler is linearly extrapolated to 1000Hz to meet the high update requirements of the baseband tracking loop. Then, the auxiliary Doppler is input to a numerically controlled oscillator (NCO) to control the loop, eliminating the influence of dynamic stress, thereby compressing the bandwidth and achieving a longer integration time. A comparison of the multi-source auxiliary Doppler before and after extrapolation is shown below. Figure 5 As shown, the Doppler information before extrapolation is consistent with the frequency of inertial navigation update. After extrapolation, acceleration information is used to update Doppler information in each inertial navigation update cycle. The multi-source assisted Doppler information output is more timely and smoother, which can better compensate for the dynamic stress in the baseband loop.

[0146] Example 3

[0147] Based on the system provided in Example 1, a four-quadrant discriminator is proposed to perform phase detection on the loop:

[0148]

[0149] In equation (15):

[0150] This indicates the output structure of the phase detector;

[0151] ( ) indicates the calculation of the arctangent function;

[0152] and are the results of coherent integration respectively.

[0153] In this embodiment, the principle of baseband tracking loop data processing is shown in Figure 6 In the preprocessing module, the intermediate frequency signal is stripped of the carrier and pseudo-random code through the mixer and correlator, and then input to the long coherent integration control module for long coherent integration. The long coherent integration output and two-way signals to the carrier phase discriminator control module. In the carrier phase discriminator control module, the phase discriminator is used to discriminate the phase of the I and Q two-way signals, and the phase discrimination error is filtered and sent to the NCO for loop control. At the same time, the carrier NCO also combines the multi-source auxiliary Doppler information from the outside for joint control. Among them, the embodiment preferably proposes to use external navigation bit assistance to eliminate the influence of navigation bit jump after successfully obtaining fixed solution and continuously positioning for more than 6 to 10 epochs, and switch to four-quadrant phase discrimination mode. As shown in Figure 7 It can be seen from the figure that the linear working range of the four-quadrant discriminator is expanded to ±180°, which is twice the discrimination range of the traditional two-quadrant discriminator (±90°). Figure 8 The comparison of two-quadrant and four-quadrant discrimination results of BD13 satellite in complex environment is shown. It can be seen that the two-quadrant discriminator loop will lose lock frequently, while the four-quadrant discriminator can maintain stable tracking of the signal.

[0154] According to the results of the four-quadrant discriminator, the signal anomaly indicator is proposed to control the extraction of the carrier phase observation value:

[0155]

[0156] In formula (16):

[0157] and respectively represent the epoch and corresponding time.

[0158] and are the phase discrimination results of the current time and the last time respectively.

[0159] When the result of the indicator exceeds the given 180° threshold, the signal anomaly flag is determined to be true, indicating that cycle slip may occur. The carrier phase observation of this epoch will be determined to be an error and will not be used for GNSS positioning.

[0160] Embodiment 4

[0161] The positioning method is realized on the basis of the system provided in the above embodiment, and is used for precise positioning and navigation of robots and autonomous vehicles in complex environments.

[0162] In order to facilitate understanding of the technical effects of the present application, experimental tests are carried out for verification:

[0163] A wheeled robot is used to test in a complex environment including boulevards, building obstructions and the like, to verify the feasibility and advancement of the present application. A high-precision positioning system (POS) Leador A15 is used as a reference true value. GNSS signals are recorded by a GNSS satellite signal recording and playback instrument Spirent GSS6450, which down-converts satellite signals to an intermediate frequency, and only signals of BD and GPS L1 frequency points are processed in the experiment. A monocular camera AVT G192 is used to collect grayscale images, and all data are synchronized to GPS time through a self-developed combined navigation module. In order to verify the effect of the multi-source deep combination system of the present application on carrier tracking and positioning output, the baseband tracking, observation and final positioning results are evaluated from three levels.

[0164] Figure 9 The loop phase discriminator outputs of BD13 satellite under different coherent integration times (including 20ms and 300ms) are compared. It can be seen that as the integration time is prolonged, the loop phase discriminator error is obviously reduced, and the tracking performance is significantly improved.

[0165] The carrier phase error is calculated by a high-precision reference system, and the carrier phase error result only includes cycle slips and white noise, so that the observation level is analyzed. Figure 10 The error conditions of the BD13 satellite output carrier observation under 20ms and 300ms are compared. It can be seen that, compared with 300ms integration time, 20ms integration time is more likely to cause a large number of half-cycle slips and whole-cycle slips; at the same time, the carrier phase error is significantly reduced as the integration time is prolonged. Long coherent integration tracking can fundamentally improve the carrier phase accuracy and continuity. Based on multi-sensor deep combination, the coherent integration time can be sufficiently prolonged to ensure accurate and continuous carrier phase observation output in extremely harsh environments.

[0166] Figure 11The horizontal positioning results of the traditional commercial receiver and the multi-source deep combination system with 300 ms long coherent integration are compared. The RTK positioning ambiguity fixing rates of the two are 7.6% and 100.0%, respectively, and the positioning errors are meter-level and centimeter-level, respectively. In complex environments, the traditional commercial receiver cannot fix the ambiguity due to frequent signal loss of lock, and the positioning error is large, while the multi-source deep combination system with 300 ms long coherent integration can maintain stable tracking of the signal carrier phase and provide centimeter-level positioning in complex environments. The positioning results show that the multi-source deep combination system has high performance in signal weakening and multipath mitigation, and solves the problem of poor availability of high-precision GNSS in complex environments.

[0167] The test results show that the GNSS / INS / Vision deep combination system based on GNSS carrier phase long coherent integration tracking of the application improves the availability of GNSS carrier phase in extremely challenging environments, and meets the navigation needs of robots and autonomous vehicles in complex environments. It is worth noting that the 300 ms long coherent integration architecture used by the multi-source deep combination system in this test example is set for the data in the complex environment tested, and does not represent the limit of the coherent integration period and the optimal sensitivity of the carrier phase of the method. When testing other data sets, the integration period can be reasonably extended according to the actual situation to achieve better signal tracking and positioning effect.

[0168] In specific implementation, the method proposed by the technical scheme of the application can be automatically run by a person skilled in the art using computer software technology, for example, providing corresponding positioning software for user operation, which can be integrated into software of a robot or an autonomous vehicle. The system device for implementing the method, such as a computer readable storage medium storing the corresponding computer program of the technical scheme of the application and a computer device including the running of the corresponding computer program, should also be within the protection scope of the application.

[0169] In a possible embodiment, the application further provides a computer program product, which includes a computer program that can be stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor, so that the computer can execute the software processing part of each method.

[0170] In a possible embodiment, the application further provides a non-transitory computer readable storage medium, which stores a computer program that is executed by a processor to implement the software processing part of each method.

[0171] The apparatus embodiments described above are only illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0172] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be implemented by means of software plus necessary universal hardware platforms, and of course can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in various embodiments or some parts of the embodiments.

[0173] The specific embodiments described herein are merely illustrative of the spirit of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or replace them with similar ways, without departing from the spirit of the present application or exceeding the scope defined by the appended claims.

Claims

1. A GNSS / INS / Vision deep integration system oriented to GNSS carrier phase long coherent integration tracking, characterized in that: The upper multi-source data fusion subsystem, the bottom GNSS baseband signal processing subsystem and the deep combination auxiliary module, The upper multi-source data fusion subsystem comprises a GNSS precise positioning module, a visual inertial odometer module and a multi-source data fusion module. The visual inertial odometer module comprises an inertial navigation module and a visual navigation module. The inertial navigation module performs inertial mechanical arrangement based on the raw data of the accelerometer and the gyroscope of the IMU, propagates the navigation state in the module, and outputs the attitude, speed and position information of the carrier. The visual navigation module extracts, tracks and triangulates feature points using camera data, and fuses the results with the inertial mechanical arrangement in the subsequent process. The multi-source data fusion module uses the GNSS positioning results from the GNSS precise positioning module and the results from the visual inertial odometer module as inputs, uses a multi-state constraint Kalman filter to fuse the multi-source data, and outputs high-precision fusion positioning results. The deep combination auxiliary module generates multi-source dynamic auxiliary Doppler information based on the high-precision fusion positioning results and satellite dynamic information, and inputs the information into the bottom GNSS baseband signal processing subsystem to eliminate the dynamic stress generated by the relative motion of the receiver-satellite in the coherent integration process. The bottom GNSS baseband signal processing subsystem provides pseudorange, Doppler and carrier phase observations for the upper multi-source data fusion subsystem to calculate the navigation state of the receiver. The bottom GNSS baseband signal processing subsystem comprises a signal preprocessing module, a long coherent integration control module and a carrier phase discrimination control module. The long coherent integration control module is configured to use a multi-stage convergence method to gradually extend the coherent integration time and compress the loop bandwidth. The carrier phase discrimination control module supports two-quadrant and four-quadrant discrimination mode switching, and the linear working range of the four-quadrant discrimination mode is ±180°. The long coherent integration control module reads the navigation bits through an external bit auxiliary interface, and performs long-time coherent integration of multiple navigation bit lengths based on the navigation bit transmission time after frame synchronization is valid. Through multi-stage convergence, the coherent integration time is gradually extended, and the loop bandwidth is compressed to achieve the target coherent integration time. The initial coherent integration time is 1 navigation bit length, and the target coherent integration time is N navigation bit lengths. A plurality of loop convergence stages are set between the initial and target coherent integration times. Loop update calculates loop lock indicator each time , set the PLI threshold , when the predetermined number of times is continuously detected , determine loop convergence, switch to the next stage until the target coherent integration time is reached; The carrier phase discrimination control module uses the same external bit auxiliary interface as the long coherent integration control module, and switches between two-quadrant and four-quadrant discrimination modes under the assistance of external bits. The initial loop uses two-quadrant discrimination mode, and after successfully obtaining a fixed solution and continuously positioning for more than a certain number of epochs, it is considered that the loop enters a stable tracking state, and the four-quadrant discrimination mode is switched to using external navigation bits.

2. The deep combining system of claim 1, wherein, The bottom GNSS baseband signal processing subsystem is configured to extract and provide the pseudorange, Doppler and carrier phase observations at an adjustable frequency ranging from 1 Hz to 50 Hz.

3. The deep combining system of claim 1, wherein, In the signal preprocessing module, after the intermediate frequency signal is stripped of carrier and pseudo-random code by the frequency mixer and correlator, I and Q channel correlation results are generated and input to the long coherent integration control module for coherent integration of the navigation bit length.

4. The deep combining system of claim 1, wherein, The carrier phase discriminator control module supports two and four quadrant phase discriminator mode switching. In the initial stage, the two quadrant phase discriminator mode is used. When the loop enters the stable tracking state, the external navigation bit is used to assist the switching to the four quadrant phase discriminator mode. The four quadrant discriminator is used to discriminate the phase of the loop as follows: In the formula: represents the output structure of the phase detector; () represents the calculation of the arctangent function; and are the coherent integration results, respectively.

5. The deep combining system of claim 1, wherein, The deep combination auxiliary module generates multi-source dynamic auxiliary Doppler information based on the receiver dynamics and satellite dynamics. The receiver dynamics is obtained by fusing the positioning results and compensating the antenna boom arm. The satellite dynamics is obtained by solving the ephemeris from the GNSS precise positioning module. The multi-source dynamic auxiliary Doppler information is input to the underlying GNSS baseband signal processing subsystem after being extrapolated by the uniform acceleration model to eliminate dynamic stress. The acceleration of the receiver motion is projected by the acceleration of the high-precision fusion positioning result to obtain: In the formula: and and denote the antenna and inertial navigation acceleration in the navigation coordinate system, respectively; denotes the direction cosine matrix from the vehicle coordinate system to the navigation coordinate system; denotes the inertial angular velocity in the carrier coordinate system; denotes the derivative of the inertial angular velocity; denotes an antenna mast arm.

6. A positioning method implemented by a GNSS / INS / Vision deep integration system using the GNSS carrier phase long coherent integration tracking according to any one of claims 1 to 5. It is used for precise positioning and navigation of robots and autonomous vehicles in complex environments.

Images