First-stage deep coupling positioning system and method based on GNSS original signal and laser radar point cloud and application
By constructing a one-stage deep-coupled positioning system based on raw GNSS signals and lidar point clouds, the positioning accuracy problem in GNSS signal obstruction environments was solved, achieving high-precision and robust positioning in complex environments.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-27
- Publication Date
- 2026-04-17
AI Technical Summary
Existing technologies lack a navigation solution that can achieve continuous, stable, and high-precision positioning in various scenarios such as open environments, urban canyons, and intermittent signals. In particular, the performance of the integrated navigation system drops sharply when GNSS signals are blocked or severely interfered with.
A one-stage deep-coupled positioning system based on raw GNSS signals and lidar point clouds is adopted. A factor graph optimization model is constructed through the data processing units of the IMU module, lidar module, and GNSS receiver module. The model integrates the IMU pre-integration residual, lidar odometry residual, pseudorange residual, and carrier phase residual factor to achieve high-precision positioning.
It provides centimeter-level accuracy when GNSS signals are available, and maintains system positioning through IMU and lidar factors when signals are lost. It is suitable for complex environments, achieves seamless degradation and recovery, and improves the robustness and accuracy of the positioning system.
Smart Images
Figure CN121876952A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of navigation and positioning technology, specifically relating to a one-stage deep coupling positioning system, method, and application based on GNSS raw signals and lidar point clouds. Background Technology
[0002] Global Navigation Satellite Systems (GNSS) provide global, all-day absolute position information and are one of the core sensors for positioning outdoor mobile platforms such as autonomous vehicles, robots, and drones. However, GNSS signals are easily blocked, reflected (multipath effect), and interfered with in environments such as urban canyons, dense forests, underground passages, or under overpasses, leading to decreased positioning accuracy or even complete failure.
[0003] To improve the continuity and reliability of positioning, the industry generally adopts a combined navigation scheme of Global Navigation Satellite System (GNSS) and Inertial Navigation System (INS). Currently, the mainstream combination methods are divided into loose combination and tight combination. Loose combination is the most widely used combined navigation method on the market, where GNSS and INS process their respective data independently. GNSS provides external position and velocity information, while INS provides continuous position, velocity, and attitude information based on internal accelerometers and gyroscopes. At the data processing level, GPS and INS data are merged, using GNSS data to correct the accumulated errors of INS, thus achieving the purpose of loose combination navigation and positioning. Tight combination, on the other hand, fuses data more closely. Tight combination directly fuses the raw GPS signals (such as pseudorange, carrier phase, and Doppler data) with the acceleration and angular velocity data of INS. Because this fusion is at the raw signal level, its real-time performance and positioning accuracy far surpass those of loose combination.
[0004] Both loose and tight navigation methods rely on GPS signals. If the GPS signal is lost or disconnected, the performance of the integrated system will be significantly affected. Therefore, it can only be used in unobstructed, open environments. If RTK technology is used, a nearby base station and a stable network connection are also required. This severely limits the application scenarios for autonomous driving and low-speed outdoor robots, failing to meet the navigation and positioning needs of urban canyons, dense forests, and remote mountainous areas.
[0005] Therefore, existing technologies lack a one-stage solution that can deeply integrate raw GNSS signals, raw IMU data, and raw data from external sensing sensors (such as lidar) and process them in real time within a unified optimization framework, thereby achieving continuous, stable, and high-precision positioning in various scenarios such as open environments, urban canyons, and intermittent signals. Summary of the Invention
[0006] To address the problems in the prior art, the present invention aims to provide [something].
[0007] To achieve the above objectives and technical effects, the technical solution adopted by this invention is as follows: A one-stage deeply coupled positioning system based on raw GNSS signals and lidar point clouds includes: The IMU module is used to acquire inertial measurement unit data; A lidar module used to acquire lidar point cloud data; A GNSS receiver module for receiving raw Global Navigation Satellite System observation data, wherein the raw GNSS observation data includes at least pseudorange observations and carrier phase observations; The data processing unit is communicatively connected to the IMU module, the lidar module, and the GNSS receiver module. The data processing unit is configured to perform the following steps: Within the sliding time window, observation data is received from the IMU module, the lidar module, and the GNSS receiver module, namely, the angular velocity and acceleration measurements detected by the IMU module, the 3D point cloud frames detected by the lidar module, and the raw pseudorange and carrier phase observations detected by the GNSS receiver module. A unified factor graph optimization model is constructed. The state variables of the factor graph optimization model include: the pose, velocity, IMU bias, and receiver clock bias of the mobile platform in the local coordinate system, as well as the transformation from the local coordinate system to the geocentric geofixed coordinate system. Based on the received observation data, corresponding residual factors are constructed in the factor graph optimization model. The residual factors include at least: lidar odometry residual factor, IMU pre-integration residual factor, pseudorange residual factor, and carrier phase residual factor based on dual-difference technology. By minimizing the weighted sum of squares of all residual factors, the state variables within the sliding time window are jointly optimized and solved to output the high-precision positioning result of the mobile platform in the geocentric-ground-fixed coordinate system. The carrier phase residual factor based on dual-difference technology is constructed by performing dual-difference calculations on the carrier phase observations of at least two different satellites at different times under the same signal frequency band, and is used to eliminate satellite clock errors, receiver clock errors and atmospheric delay errors.
[0008] Furthermore, the lidar odometry residual factor is constructed in the following manner: The current frame's LiDAR point cloud is registered with the local sub-image using the iterative nearest point algorithm to obtain the inter-frame relative pose transformation. This relative pose transformation is then compared with the estimated state variable pose at the corresponding time to construct the pose residual.
[0009] Furthermore, the IMU pre-integration residual factor is calculated based on the pre-integration of the angular velocity and acceleration measurements of the IMU between two consecutive state times. Its residual terms include rotation residual, velocity residual and position residual. The accelerometer bias and gyroscope bias of the IMU are modeled as a random walk process that changes slowly over time, and corresponding bias residual factors are constructed.
[0010] Furthermore, the pseudorange residual factor is constructed in the following manner: Based on the ephemeris data of GNSS satellites, the position of the satellite in the geocentric-ground-fixed coordinate system at the time of signal transmission is calculated. Combined with the estimated position of the mobile platform antenna, the receiver clock error, and the modeled tropospheric and ionospheric delays, a pseudorange estimate is calculated. This estimate is then compared with the pseudorange value actually observed by the GNSS receiver module to construct the residual.
[0011] Furthermore, the carrier phase residual factor based on the dual-difference technique is constructed in the following way: Select the satellite with the longest continuous visibility within the time window as the reference satellite; For each other satellite in the same signal frequency band, form a satellite pair with the reference satellite; For each satellite pair, select two different observation times; Calculate the difference between the carrier phase observations of the two satellites in the satellite pair at the first observation time. Calculate the difference between the carrier phase observations of the two satellites in the satellite pair at the second observation time. The two single-difference observations are then differentiald to obtain the double-difference carrier phase observations; Based on the estimated positions of the mobile platform at the two observation times and the positions of the two satellites at the two observation times, the theoretical geometric distance difference of the dual differential carrier phase is calculated. The theoretical geometric distance difference is compared with the dual-differential carrier phase observation to construct the residual.
[0012] Furthermore, the data processing unit is configured as follows: When GNSS signals are partially lost or degraded in quality, the system's positioning function is maintained by the continuous IMU pre-integration residual factor and lidar odometry residual factor in the factor graph optimization model. When the GNSS signal is recovered, the newly acquired pseudorange residual factor and carrier phase residual factor are automatically incorporated into the optimization process to achieve seamless integration.
[0013] This invention also discloses a one-stage deep-coupled positioning method based on raw GNSS signals and lidar point clouds. Positioning is performed using a one-stage deep-coupled positioning system based on raw GNSS signals and lidar point clouds as described above, and includes the following steps: Within the sliding time window, observation data is received from the IMU module, lidar module, and GNSS receiver module; A unified factor graph optimization model is constructed. The state variables of the factor graph optimization model include: the pose, velocity, IMU bias, and receiver clock bias of the mobile platform in the local coordinate system, as well as the transformation from the local coordinate system to the geocentric geofixed coordinate system. Based on the received observation data, corresponding residual factors are constructed in the factor graph optimization model. The residual factors include at least: lidar odometry residual factor, IMU pre-integration residual factor, pseudorange residual factor, and carrier phase residual factor based on dual-difference technology. By minimizing the weighted sum of squares of all residual factors, the state variables within the sliding time window are jointly optimized and solved to output the high-precision positioning result of the mobile platform in the geocentric coordinate system.
[0014] Furthermore, the carrier phase residual factor based on dual-difference technology is constructed by performing dual-difference calculations on the carrier phase observations of at least two different satellites at different times under the same signal frequency band, and is used to eliminate satellite clock errors, receiver clock errors and atmospheric delay errors.
[0015] Furthermore, the method also includes: When GNSS signals are partially lost or degraded in quality, the system's positioning function is maintained by the continuous IMU pre-integration residual factor and lidar odometry residual factor in the factor graph optimization model. When the GNSS signal is recovered, the newly acquired pseudorange residual factor and carrier phase residual factor are automatically incorporated into the optimization process to achieve seamless integration.
[0016] This invention also discloses the application of a one-stage deep coupling positioning system based on GNSS raw signals and lidar point clouds in autonomous vehicles, robots, or drones.
[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1) One-stage deep fusion: It abandons the traditional two-stage or multi-sensor loosely coupled architecture, and processes the constraints of all raw observation data synchronously in an optimization problem, avoiding mode switching and information loss, and realizing a deeper level of sensor fusion. 2) High precision and high robustness: By introducing carrier phase double differential residual factors, when GNSS signals are available, the system is provided with strong geometric constraints with centimeter-level accuracy, which greatly improves the absolute positioning accuracy; when GNSS signals are partially or completely lost, the IMU and lidar factors can maintain the system's relative positioning and short-term accuracy, achieving seamless degradation and recovery. 3) Wide environmental adaptability: It is particularly suitable for complex scenarios with intermittent GNSS signals and severe multipath effects, such as urban canyons, tree-lined roads, and under overpasses, solving the problem of the sharp performance degradation of existing integrated navigation systems in such environments; 4) Maximizing the advantages of tight coupling: Directly using raw GNSS observations (pseudorange, carrier phase) instead of calculated position information allows for more efficient use of GNSS receiver information. Even when the number of visible satellites is small, usable positioning solutions can be obtained through deep coupling with other sensors. Attached Figure Description
[0018] Figure 1 This is a schematic diagram of the coordinate system of the present invention; Figure 2 This is a schematic diagram of the factor graph of the present invention. Detailed Implementation
[0019] The present invention will now be described in detail so that its advantages and features can be more easily understood by those skilled in the art, thereby providing a clearer and more explicit definition of the scope of protection of the present invention.
[0020] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form to prepare for the more detailed descriptions that follow.
[0021] like Figure 1-2 As shown, this invention discloses a one-stage deeply coupled positioning system based on raw GNSS signals and lidar point clouds, comprising: The IMU module is used to acquire inertial measurement unit data; A lidar module used to acquire lidar point cloud data; A GNSS receiver module for receiving raw global navigation satellite system observation data, which includes at least pseudorange and carrier phase observations; The data processing unit communicates with the IMU module, lidar module, and GNSS receiver module. The data processing unit is configured to perform the following steps: Within the sliding time window, observation data is received from the IMU module, the lidar module, and the GNSS receiver module, namely, the angular velocity and acceleration measurements detected by the IMU module, the 3D point cloud frames detected by the lidar module, and the raw pseudorange and carrier phase observations detected by the GNSS receiver module. A unified factor graph optimization model is constructed. The state variables of the factor graph optimization model include: the pose, velocity, IMU bias, and receiver clock bias of the mobile platform in the local coordinate system, as well as the transformation from the local coordinate system to the geocentric geofixed coordinate system. Based on the received observation data, the corresponding residual factors are constructed in the factor graph optimization model. The residual factors include at least: lidar odometry residual factor, IMU pre-integration residual factor, pseudorange residual factor and carrier phase residual factor based on double difference technology. By minimizing the weighted sum of squares of all residual factors, the state variables within the sliding time window are jointly optimized and solved to output the high-precision positioning results of the mobile platform in the geocentric-ground-fixed coordinate system. Among them, the carrier phase residual factor based on double difference technology is constructed by performing double difference calculations on the carrier phase observation values of at least two different satellites at different times under the same signal frequency band, and is used to eliminate satellite clock error, receiver clock error and atmospheric delay error.
[0022] Before introducing the above system, we must first determine the coordinate system. Ultimately, we want a geographic coordinate system, i.e., positioning coordinates in a latitude and longitude coordinate system. However, for computational convenience, we internally use a Cartesian geocentric coordinate system E as the system's global coordinate system, with its z-axis pointing to the Earth's rotation axis and its x-axis pointing to the prime meridian. Because this system not only calculates global coordinates but also needs to maintain a continuous and smooth local trajectory during GNSS convergence, a local coordinate system W also needs to be established. The pose of this local coordinate system coincides with the platform's initial pose. The coordinate systems of each sensor on the hardware platform are defined as follows: the coordinate system B, fixed to the platform, serves as the coordinate system for the GNSS antenna; the IMU coordinate system I and the lidar coordinate system L are as follows... Figure 1 As shown.
[0023] The system hardware platform of this invention includes: a GNSS receiver (supporting multiple frequencies and systems, outputting raw observation values), an IMU (six-axis or nine-axis), a lidar, and a data processing unit responsible for core calculations (such as an industrial control computer or an embedded high-performance computing platform). Each sensor is synchronized via hardware synchronization or software timestamp alignment to ensure spatiotemporal synchronization of data.
[0024] System measurement values (input): Two time points in IMU coordinate system I and All measurements between LiDAR point cloud The pseudorange received by the GNSS receiver and dual differential signals Therefore, the input to this system is all the measurements taken within a time window: System state estimation (output): The system estimates its historical states using maximum likelihood estimation of the observed (input) data. Transform it into a computationally optimizable form: Each term in the formula corresponds to a residual component of a specific factor type and is weighted by the covariance matrix. The residuals include priors, IMU pre-integration residuals, lidar odometry residuals, and residuals from the two raw GNSS signals.
[0025] Factor diagram as follows Figure 2 As shown, the estimator is the larger circle, where, To estimate the state, the transformation The transformation from the local coordinate system W to the geocentric coordinate system E is represented by smaller colored circles, which indicate factor residual nodes, including a priori residual factors, lidar odometer residual factors, IMU pre-integration residual factors, pseudorange residual factors, and carrier phase residual factors based on dual-difference technology.
[0026] In some implementations, the lidar odometry residual factor is constructed in the following way: The current frame's LiDAR point cloud is registered with a local sub-image using an iterative nearest-point algorithm to obtain the inter-frame relative pose transformation. This relative pose transformation is then compared with the estimated state variable pose at the corresponding time moment to construct the pose residual. The lidar is registered to a local sub-graph using the ICP algorithm, and the odometry result is output as time. and The relative pose factor is represented using the following residuals: in, The estimated pose. Estimated using the ICP algorithm.
[0027] In some implementations, the IMU pre-integration residual factor is calculated based on the pre-integration of angular velocity and acceleration measurements of the IMU between two consecutive state times, and its residual terms include rotational residual, velocity residual, and position residual; the IMU's accelerometer bias and gyroscope bias are modeled as a slowly varying random walk process over time, and corresponding bias residual factors are constructed: Using the common method of IMU measurement processing in academia and industry, "IMU pre-integration," a residual factor for IMU pre-integration is constructed: in: When dealing with the IMU bias, the biases of the accelerometer and gyroscope are modeled as quantities that change slowly over time. Brownian motion is used to model the two biases, i.e., integrated white noise: At two consecutive time points and Integrating between them, we get: in, and They are two consecutive terms with a mean of 0 and a covariance of . as well as The noise function. Therefore, for the two biases of the IMU, residual factors can be constructed: .
[0028] In some implementations, the pseudorange residual factor is constructed in the following way: Based on GNSS satellite ephemeris data, the satellite's position in the geocentric-ground-fixed coordinate system at the time of signal transmission is calculated. Combined with the estimated mobile platform antenna position, receiver clock bias, and modeled tropospheric and ionospheric delays, a pseudorange estimate is obtained. This estimate is then compared with the pseudorange value actually observed by the GNSS receiver module to construct the residual. Regarding time Each satellite signal acquired Each GNSS receiver will receive a pseudorange Pseudorange is obtained by multiplying the observed signal time by the speed of light. The pseudorange result approximates the actual satellite-receiver distance in space, but it is affected by other factors, such as signal delays caused by the troposphere and ionosphere, and clock offsets used to determine signal transmission time. The pseudorange residual factor for a single satellite signal obtained from the receiver can be approximated as: in Indicates the time of observation The corresponding geocentric coordinate system at the time of signal transmission The satellite's 3D position is determined and corrected for by the Earth's rotation. Indicates the spatial signal delay in the troposphere. This indicates the signal delay in the ionosphere. Representing the speed of light, the relative offset between the receiver and the GNSS clock is... Adjust pseudorange To correct satellite clock deviation and relativity Antenna position and clock offset are the state variables we need to estimate. If a multi-band receiver is used, a satellite can have multiple residuals (one residual for each band of the signal received by the receiver). Simultaneously, observations from different bands can be combined to estimate atmospheric delay.
[0029] In some implementations, the carrier phase residual factor based on the double-difference technique is constructed in the following way: Select the satellite with the longest continuous visibility within the time window as the reference satellite; For each other satellite in the same signal frequency band, form a satellite pair with the reference satellite; For each satellite pair, select two different observation times; Calculate the difference between the carrier phase observations of the two satellites in the satellite pair at the first observation time. Calculate the difference between the carrier phase observations of the two satellites in the satellite pair at the second observation time. The two single-difference observations are then differentiald to obtain the double-difference carrier phase observations; Based on the estimated positions of the mobile platform at the two observation times and the positions of the two satellites at the two observation times, the theoretical geometric distance difference of the dual differential carrier phase is calculated. The theoretical geometric distance difference is compared with the dual-differential carrier phase observation to construct the residual: The residual factor of the carrier phase is similar to that of the pseudorange residual, and can be written as: In the formula, most of the variables have the same meaning as those in the pseudo-range residual factor equation. This refers to the clockwork effect caused by the interaction between the satellite's orientation and the circularly polarized carrier wave. This indicates the wavelength of the circularly polarized carrier wave. This indicates an unknown wavelength offset. The advantage of carrier phase observation over pseudorange observation is its measurement noise of only 5 mm. However, accurately modeling satellite position and atmospheric delay in real-time is challenging because these are not known with sub-meter accuracy. Furthermore, there is an unknown quantity that needs to be estimated: integer ambiguity. To address this problem, this invention proposes a double-difference technique, which combines multiple observations to cancel out unknown or inaccurately known terms. Specifically, the double-difference technique combines the current time... Two observations of two different satellites within the same Global Navigation Satellite System (GNSS) and signal frequency band, and observations conducted earlier than... At some point This involves combining two sets of older observations from the same satellites. This data combination is used to improve the accuracy and reliability of navigation solutions. The following is a calculation formula that uses the double-difference technique to combine multiple observations to eliminate uncertainties: in, It is a double-difference observation, and its calculation is based on the differences of four observations, where Carrier phase Carrier phase observations after satellite clock deviation and relativistic correction. If time... and satellite signals between and The atmospheric delay is negligible, and the signal is always successfully locked, i.e. , The residual is: For each pair of bands and satellite systems (GPS, GLONASS, Galileo, BeiDou), we select the first satellite signal. The satellite with the longest continuous visibility. Regarding the signal from the second satellite. The remaining satellite observations are iterated over within the same signal bandwidth. In this way, a residual is created for each satellite signal pair.
[0030] The system in time The state can be represented as: in, , These represent the rotation and translation of coordinate system B relative to the world coordinate system W, respectively. This represents the linear velocity of coordinate system B relative to world coordinate system W. These represent the angular velocity offset and acceleration offset on the IMU body coordinate system I, respectively. This is a vector representing the clock offset between each satellite system and the receiver clock. Because the system includes a lidar system, it is also necessary to estimate the motion from the world coordinate system W to the geocentric Earth-fixed coordinate system E. Therefore, the unknown quantities of all historical states are: in, Indicates the time All states within a fixed-length sliding window.
[0031] In some implementations, the data processing unit is configured to: When GNSS signals are partially lost or degraded in quality, the system's positioning function is maintained by the continuous IMU pre-integration residual factor and lidar odometry residual factor in the factor graph optimization model. When the GNSS signal is recovered, the newly acquired pseudorange residual factor and carrier phase residual factor are automatically incorporated into the optimization process to achieve seamless integration.
[0032] This invention also discloses a one-stage deep-coupled positioning method based on raw GNSS signals and lidar point clouds. Positioning is performed using a one-stage deep-coupled positioning system based on raw GNSS signals and lidar point clouds as described above, and includes the following steps: Within the sliding time window, observation data is received from the IMU module, lidar module, and GNSS receiver module; A unified factor graph optimization model is constructed. The state variables of the factor graph optimization model include: the pose, velocity, IMU bias, and receiver clock bias of the mobile platform in the local coordinate system, as well as the transformation from the local coordinate system to the geocentric geofixed coordinate system. Based on the received observation data, the corresponding residual factors are constructed in the factor graph optimization model. The residual factors include at least: lidar odometry residual factor, IMU pre-integration residual factor, pseudorange residual factor and carrier phase residual factor based on double difference technology. By minimizing the weighted sum of squares of all residual factors, the state variables within the sliding time window are jointly optimized to output the high-precision positioning results of the mobile platform in the geocentric-ground-fixed coordinate system.
[0033] In some embodiments, the above method further includes: When GNSS signals are partially lost or degraded in quality, the system's positioning function is maintained by the continuous IMU pre-integration residual factor and lidar odometry residual factor in the factor graph optimization model. When the GNSS signal is recovered, the newly acquired pseudorange residual factor and carrier phase residual factor are automatically incorporated into the optimization process to achieve seamless integration.
[0034] In some implementations, the carrier phase residual factor based on the dual-difference technique is constructed by performing dual-difference calculations on the carrier phase observations of at least two different satellites at different times under the same signal frequency band, and is used to eliminate satellite clock errors, receiver clock errors and atmospheric delay errors.
[0035] This invention also discloses the application of a one-stage deep coupling positioning system based on GNSS raw signals and lidar point clouds in autonomous vehicles, robots, or drones.
[0036] Any parts or structures not specifically described in this invention can be made using existing technologies or products, and will not be elaborated upon here.
[0037] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A one-stage deep-coupled positioning system based on GNSS raw signals and LiDAR point clouds, characterized in that, include: The IMU module is used to acquire inertial measurement unit data; A lidar module used to acquire lidar point cloud data; A GNSS receiver module for receiving raw Global Navigation Satellite System observation data, wherein the raw GNSS observation data includes at least pseudorange observations and carrier phase observations; The data processing unit is communicatively connected to the IMU module, the lidar module, and the GNSS receiver module. The data processing unit is configured to perform the following steps: Within the sliding time window, observation data is received from the IMU module, lidar module, and GNSS receiver module; A unified factor graph optimization model is constructed. The state variables of the factor graph optimization model include: the pose, velocity, IMU bias, and receiver clock bias of the mobile platform in the local coordinate system, as well as the transformation from the local coordinate system to the geocentric geofixed coordinate system. Based on the received observation data, corresponding residual factors are constructed in the factor graph optimization model. The residual factors include at least: lidar odometry residual factor, IMU pre-integration residual factor, pseudorange residual factor, and carrier phase residual factor based on dual-difference technology. By minimizing the weighted sum of squares of all residual factors, the state variables within the sliding time window are jointly optimized and solved to output the high-precision positioning result of the mobile platform in the geocentric-ground-fixed coordinate system. The carrier phase residual factor based on dual-difference technology is constructed by performing dual-difference calculations on the carrier phase observations of at least two different satellites at different times under the same signal frequency band, and is used to eliminate satellite clock errors, receiver clock errors and atmospheric delay errors.
2. The one-stage deep-coupled positioning system based on GNSS raw signals and lidar point clouds according to claim 1, characterized in that, The lidar odometry residual factor is constructed in the following way: The current frame's LiDAR point cloud is registered with the local sub-image using the iterative nearest point algorithm to obtain the inter-frame relative pose transformation. This relative pose transformation is then compared with the estimated state variable pose at the corresponding time to construct the pose residual.
3. The one-stage deep-coupled positioning system based on GNSS raw signals and lidar point clouds according to claim 1, characterized in that, The IMU pre-integration residual factor is calculated based on the pre-integration of the angular velocity and acceleration measurements of the IMU between two consecutive state times. Its residual terms include rotation residual, velocity residual and position residual. The accelerometer bias and gyroscope bias of the IMU are modeled as a random walk process that changes slowly over time, and corresponding bias residual factors are constructed.
4. The one-stage deep coupling positioning system based on GNSS raw signals and lidar point clouds according to claim 1, characterized in that, The pseudo-range residual factor is constructed in the following way: Based on the ephemeris data of GNSS satellites, the position of the satellite in the geocentric-ground-fixed coordinate system at the time of signal transmission is calculated. Combined with the estimated position of the mobile platform antenna, the receiver clock error, and the modeled tropospheric and ionospheric delays, a pseudorange estimate is calculated. This estimate is then compared with the pseudorange value actually observed by the GNSS receiver module to construct the residual.
5. A one-stage deep-coupled positioning system based on GNSS raw signals and lidar point clouds according to claim 1, characterized in that, The carrier phase residual factor based on double-difference technology is constructed in the following way: Select the satellite with the longest continuous visibility within the time window as the reference satellite; For each other satellite in the same signal frequency band, form a satellite pair with the reference satellite; For each satellite pair, select two different observation times; Calculate the difference between the carrier phase observations of the two satellites in the satellite pair at the first observation time. Calculate the difference between the carrier phase observations of the two satellites in the satellite pair at the second observation time. The two single-difference observations are then differentiald to obtain the double-difference carrier phase observations; Based on the estimated positions of the mobile platform at the two observation times and the positions of the two satellites at the two observation times, the theoretical geometric distance difference of the dual differential carrier phase is calculated. The theoretical geometric distance difference is compared with the dual-differential carrier phase observation to construct the residual.
6. The one-stage deep coupling positioning system based on GNSS raw signals and lidar point clouds according to claim 1, characterized in that, The data processing unit is configured as follows: When GNSS signals are partially lost or degraded in quality, the positioning function of the system is maintained by the continuous IMU pre-integration residual factor and lidar odometry residual factor in the factor graph optimization model. When the GNSS signal is recovered, the newly acquired pseudorange residual factor and carrier phase residual factor are automatically incorporated into the optimization process to achieve seamless integration.
7. A one-stage deep coupling positioning method based on raw GNSS signals and lidar point clouds, characterized in that, Positioning using a one-stage deep-coupled positioning system based on raw GNSS signals and lidar point clouds as described in any one of claims 1-6 includes the following steps: Within the sliding time window, observation data is received from the IMU module, lidar module, and GNSS receiver module; A unified factor graph optimization model is constructed. The state variables of the factor graph optimization model include: the pose, velocity, IMU bias, and receiver clock bias of the mobile platform in the local coordinate system, as well as the transformation from the local coordinate system to the geocentric geofixed coordinate system. Based on the received observation data, corresponding residual factors are constructed in the factor graph optimization model. The residual factors include at least: lidar odometry residual factor, IMU pre-integration residual factor, pseudorange residual factor, and carrier phase residual factor based on dual-difference technology. By minimizing the weighted sum of squares of all residual factors, the state variables within the sliding time window are jointly optimized and solved to output the high-precision positioning result of the mobile platform in the geocentric coordinate system.
8. The one-stage deep coupling positioning method based on GNSS raw signals and lidar point clouds according to claim 7, characterized in that, The carrier phase residual factor based on dual-difference technology is constructed by performing dual-difference calculations on the carrier phase observations of at least two different satellites at different times under the same signal frequency band. It is used to eliminate satellite clock errors, receiver clock errors and atmospheric delay errors.
9. The one-stage deep coupling positioning method based on GNSS raw signals and lidar point clouds according to claim 7, characterized in that, The method further includes: When GNSS signals are partially lost or degraded in quality, the positioning function of the system is maintained by the continuous IMU pre-integration residual factor and lidar odometry residual factor in the factor graph optimization model. When the GNSS signal is recovered, the newly acquired pseudorange residual factor and carrier phase residual factor are automatically incorporated into the optimization process to achieve seamless integration.
10. The application of a one-stage deep-coupled positioning system based on raw GNSS signals and lidar point clouds as described in any one of claims 1-6 in autonomous vehicles, robots, or drones.