Port gantry crane GNSS lane deviation automatic calibration and correction method
By fitting the deviation relationship between the lane line information detected by the camera and the GNSS receiver, and using a linear model and tilt compensation technology, the problem of tedious and limited accuracy of manual calibration of GNSS receivers on port gantry cranes was solved, and high-precision automatic lane correction and calibration were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-07
- Publication Date
- 2026-04-14
AI Technical Summary
In existing technologies, GNSS receivers on port gantry cranes suffer from system deviations, and manual calibration is cumbersome and has limited accuracy, making it difficult to meet high-precision requirements.
Automatic lane correction is achieved by fitting the deviation relationship between the lane line information detected by the camera and the GNSS receiver. Linear model fitting and tilt compensation technology are used to automatically calibrate the GNSS lane deviation.
It achieves high-precision automatic calibration and correction of GNSS lane deviation, reduces manual intervention, improves calibration efficiency and accuracy, supports fully automatic and manual calibration modes, adapts to complex working conditions, and ensures the continuity of lane deviation output.
Smart Images
Figure CN121849792A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of lifting system transportation, and in particular to a method for automatic calibration and correction of GNSS lane deviation of port gantry cranes. Background Technology
[0002] In automated container terminal yards, rubber-tired gantry cranes (RTGs) move along fixed lanes between different yards and bays. To achieve automated movement and precise alignment, GNSS receivers are typically installed on the RTGs to obtain global position, while cameras are installed below the gantry to detect ground lane lines as alignment references.
[0003] However, existing GNSS (Global Navigation Satellite System) receivers suffer from systematic bias, and manual calibration is cumbersome and has limited accuracy. Due to antenna installation errors, coordinate transformation errors, multipath effects, etc., there is a stable systematic bias between the original coordinates of the GNSS receiver and the lane centerline. Currently, calibration is mostly performed manually by setting up points in the yard and repeatedly adjusting parameters, which is labor-intensive, error-prone, and the results depend on personal experience. In actual production, the average lane deviation of manual calibration is usually around 10cm, which is insufficient to meet higher accuracy requirements.
[0004] Therefore, it is necessary to provide an automatic calibration and correction method for GNSS lane deviation of port gantry cranes to solve the above problems. Summary of the Invention
[0005] This application provides a method for automatic calibration and correction of GNSS lane deviation for port gantry cranes. By fitting the lane line information detected by the camera and the deviation relationship between the GNSS receiver and the lane centerline, automatic lane correction is achieved.
[0006] This application provides a method for automatic calibration and correction of GNSS lane deviation for port gantry cranes, the method comprising the following steps: Acquire raw data from the GNSS receiver, including gnssX, gnssY, angle, and status; Acquire lane line attitude data output by the camera, the lane line attitude data including camera lane deviation and detection status; Obtain the position and attitude of the RTG in the yard coordinate system, wherein the position includes the yard and bay position, and the attitude includes pitch angle, roll angle and gantry height; Based on the timestamp, the raw data from the GNSS receiver and the lane line attitude data output by the camera are matched within a preset time period; The data of the stockpile and bay position are read, and the data of the absolute value of the offset between the edge bay position and the center of the stockpile that exceeds the first threshold are filtered to obtain multiple sets of valid calibration samples. Each set of data in the multiple sets of valid calibration samples includes gnssX, gnssY, laneDevFromCamera, yard, bay, and timestamp. Obtain the Euler angles and Euclidean transformations corresponding to the stockpile and the position, and calculate the ground displacement deviation of the GNSS receiver in the trolley direction and the gantry direction based on the gantry height; During the fitting phase, the lane deviation value provided by the camera is added to the tilt compensation amount so that the fitting parameters reflect the geometric relationship under the tilt-free state; during the real-time deviation calculation phase, the tilt compensation amount is subtracted from the predicted deviation. For each storage yard and each GNSS receiver, calibration samples aggregated by position are collected. Once a preset value is reached, fitting is triggered. A linear model is fitted using GNSS coordinates and tilt-compensated lane deviation as inputs.
[0007] Preferably, the tilt-compensated lane deviation is calculated using the following formula:
[0008] in, For lane deviation after tilt compensation, The geometric center of the sample coordinates is used for coordinate translation. For unit normal vector components, This is the intercept term.
[0009] Preferably, the linear model is fitted using the following method: When the data linearity is high and the noise is low, use a band... The least squares fitting method with constraints ensures that the normal vector is of unit length; When there are few data points or the distribution is irregular, a PCA-based principal axis / normal direction fitting method is adopted.
[0010] Preferably, when a preset condition is met, this set of parameters is recorded as valid calibration parameters, whereby the preset condition is: .
[0011] Preferably, for the trajectory points of multiple bays within the same stockyard, the average coordinates are calculated according to the bay and projected onto the main axis of the lane. A mapping between the bay number and the main axis coordinates is established through linear regression to estimate the spatial location of the bay at the stockyard boundary.
[0012] Preferably, if the spatial distance between two storage yards is less than a second threshold, they are determined to be physically adjacent storage yards. Based on the boundary points of the two storage yards and their respective lane models, a lane deviation model across storage yards is constructed for RTG deviation calculation in the cross-storage yard area.
[0013] Preferably, during normal RTG operation, the system receives raw data from the GNSS receiver in real time and searches for the calibration parameters of the current stockpile and corresponding equipment for each piece of raw data: if the position is within the legal range of the stockpile, the real-time lane deviation is calculated using the calibration model within the stockpile; if the position exceeds the legal range and there is a cross-stockpile correction model, the lane deviation model across the stockpile is called to calculate the deviation.
[0014] Preferably, the virtual lane deviation on the opposite side is generated using the following formula:
[0015] in, For the virtual lane deviation on the opposite side, For the actual side lane deviation, For the current gantry angle, This represents the current gantry length.
[0016] Preferably, the linear parameters of the equipment and the cross-farm parameters fitted from the multiple sets of valid calibration samples for each farm are serialized into JSON format and stored in the local file system.
[0017] This application offers the following advantages over existing technologies: It provides an automatic GNSS lane deviation calibration and correction method for port gantry cranes. The method includes the following steps: acquiring raw data from a GNSS receiver, including gnssX, gnssY, angle, and status; acquiring lane line attitude data output by a camera, including camera lane deviation and detection status; acquiring the position and attitude of the RTG in the yard coordinate system, including the yard and bay position, and the attitude including pitch angle, roll angle, and gantry height; matching the raw data from the GNSS receiver and the lane line attitude data output by the camera within a preset time period based on a timestamp; reading the yard and bay position, filtering data where the absolute value of the offset between the yard edge bay position and the center exceeds a first threshold, and obtaining multiple sets of valid calibration samples. Each set of data in the multiple sets of valid calibration samples includes gnssX, gnssY, laneDevFromCamera, yard, and... Bay and timestamp; Obtain the Euler angles and Euclidean transformations corresponding to the bay and timestamp of the stockpile; Calculate the ground displacement deviation of the GNSS receiver in the trolley direction and gantry direction based on the gantry height; In the fitting stage, add the tilt compensation amount to the lane deviation value provided by the camera so that the fitting parameters reflect the geometric relationship under the no-tilt state; In the real-time deviation calculation stage, subtract the tilt compensation amount from the predicted deviation; For each stockpile and each GNSS receiver, collect calibration samples aggregated by bay and trigger fitting after reaching a preset value; Using GNSS coordinates and the tilt-compensated lane deviation as input, fit a linear model; By fitting the lane line information detected by the camera and the deviation relationship between the GNSS receiver and the lane centerline, automatic lane correction is achieved; Furthermore, the introduction of centralization The linear fitting model with constraints and tilt compensation restricts the mapping between the GNSS receiver and the lane centerline to physically reasonable rotation and translation, eliminating scaling errors. Furthermore, the calibration data can be automatically collected using the lane line detection results during normal RTG operation, without the need for additional point deployment and dedicated debugging time; Furthermore, it supports fully automatic timed calibration, batch calibration by stockpile, and manual trigger calibration, with multiple modes that can be flexibly combined; Furthermore, by combining attitude sensors, the influence of pitch / roll angle on the ground projection of the GNSS receiver is explicitly modeled, enabling stable output of high-precision lane deviation even under complex working conditions such as changes in spreader height, uneven ground, and off-center loading. Furthermore, even if a GNSS receiver is installed on only one side of the gantry or if one side of the GNSS receiver fails, the virtual lane deviation data on the other side can be calculated in real time based on the gantry length and attitude angle, providing the upper control system with input equivalent to dual-side GNSS. Furthermore, for adjacent storage yards within the same channel, a geometric interpolation method based on boundary position coordinates and their respective deviation models is used to construct a transition deviation model across storage yards; when the RTG moves from one storage yard to another, the lane deviation output is smooth and continuous, and there will be no sudden jump in error due to storage yard switching. Attached Figure Description
[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0019] Figure 1 This is a flowchart illustrating an automatic calibration and correction method for GNSS lane deviation of a port gantry crane according to an embodiment of this application.
[0020] The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation
[0021] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.
[0022] To address the aforementioned issues, this application provides an embodiment of an automatic GNSS lane deviation calibration and correction method for port gantry cranes. This method achieves automatic lane deviation correction by fitting the lane line information detected by the camera with the deviation relationship between the GNSS receiver and the lane centerline.
[0023] Figure 1 This is a flowchart illustrating an automatic GNSS lane deviation calibration and correction method for a port gantry crane according to an embodiment of this application. Now refer to... Figure 1 This invention provides a method for automatic calibration and correction of GNSS lane deviation for port gantry cranes, the method comprising the following steps: Step S101: Obtain raw data from the GNSS receiver, including gnssX, gnssY, angle, and status; Step S102: Obtain lane line attitude data output by the camera, wherein the lane line attitude data includes camera lane deviation and detection status; Step S103: Obtain the position and attitude of the RTG in the yard coordinate system, wherein the position includes the yard and the bay, and the attitude includes pitch angle, roll angle and gantry height; Step S104: Based on the timestamp, match the raw data of the GNSS receiver and the lane attitude data output by the camera within a preset time period; Step S105: Read the stockpile and bay position, filter the data whose absolute value of the offset between the edge bay position and the center exceeds the first threshold, and obtain multiple sets of valid calibration samples. Each set of data in the multiple sets of valid calibration samples includes gnssX, gnssY, laneDevFromCamera, yard, bay, and timestamp. Step S106: Obtain the Euler angles and Euclidean transformations corresponding to the stockpile and the position, and calculate the ground displacement deviation of the GNSS receiver in the trolley direction and the gantry direction based on the gantry height; Step S017: In the fitting stage, the lane deviation value provided by the camera is added to the tilt compensation amount so that the fitting parameters reflect the geometric relationship under the no-tilt state; in the real-time deviation calculation stage, the tilt compensation amount is subtracted from the predicted deviation. Step S108: For each storage yard and each GNSS receiver, collect calibration samples aggregated by position. Once the preset value is reached, trigger fitting. Use GNSS coordinates and lane deviation after tilt compensation as input to fit a linear model.
[0024] Specifically, in step S101, gnssX and gnssY are the planar coordinate fields output by the GNSS receiver, representing the X and Y coordinates of the RTG in the yard plane coordinate system, respectively, in meters (m). The angle represents the heading / yaw angle of the RTG's travel direction or vehicle orientation, in degrees (°) or radians. The status indicates the GNSS receiver's solution quality. During the calibration data acquisition phase, valid GNSS receiver data needs to be filtered, and data indicating a fixed solution is used for calibration. During the real-time output phase, the status includes positioning validity / invalidity, for upper-level control systems or safety strategies to determine.
[0025] In step S102, the camera lane deviation is the distance in the lateral / normal direction of the lane centerline detected by the camera relative to the camera mounting reference point. The unit is meters (m). The detection status is a flag indicating the validity of the camera's lane line detection, used to discard data frames that do not detect lane lines or have insufficient confidence. When in the calibration data acquisition stage, if the detection status indicates that no lane line has been detected, the data will not participate in matching and calibration, thereby ensuring sample quality.
[0026] In step S103, the pitch angle and roll angle are used to calculate the tilt compensation. This is to compensate for the horizontal offset of the GNSS receiver mounting point's projection on the ground caused by the gantry tilt, which affects lane deviation. The gantry height is read, and the horizontal displacement component caused by the tilt angle is obtained based on the attitude angle.
[0027] The sign of the displacement component is corrected according to the installation position of the GNSS receiver to obtain the tilt compensation amount for calibration and real-time calculation.
[0028] In step S105, the edge bay is a set of bays within a preset range from the boundary bay, for example, the preset range can be 3 bays before and after. gnssX and gnssY represent the planar coordinates of the GNSS receiver, laneDevFromCamera represents the camera lane deviation, yard represents the yard number, bay represents the bay number, and timestamp represents the timestamp.
[0029] Specifically, each set of data in the multiple sets of valid calibration samples also includes gnssDeviceIndex and cameraDeviceIndex, where gnssDeviceIndex represents the device index of the gnss receiver and cameraDeviceIndex represents the device index of the camera.
[0030] In step S106, the ground displacement deviation is a compensation term introduced by the ground projection displacement of the GNSS receiver installation point caused by the gantry tilt.
[0031] In step S107, the tilt compensation amount is denoted as , is the compensation term caused by attitude, and the fitting input is: Real-time output: .
[0032] In practice, the lane deviation after tilt compensation is calculated using the following formula:
[0033] in, For lane deviation after tilt compensation, The geometric center of the sample coordinates is used for coordinate translation. For unit normal vector components, This is the intercept term. Used for centralized processing; Let be the unit normal vector component in the direction of lane deviation, satisfying the following constraints: ; This is the intercept term (unit: m). In practice, the linear model is fitted using the following method: When the data linearity is high and the noise is low, use a band... The least squares fitting method with constraints ensures that the normal vector is of unit length.
[0034] Specifically, this is determined by evaluating data characteristics using metrics, such as the number of sample points. ; Eigenvalue ratio / Aspect ratio Linearity index Lane deviation noise level .when Larger and high, When low, use belt Least squares fitting under constraints; when When the data is small or the data distribution is irregular / noisy, a PCA-based principal axis / normal axis fitting method is used.
[0035] When there are few data points or the distribution is irregular, a PCA-based principal axis / normal direction fitting method is adopted.
[0036] Specifically, configurable criteria can be used, for example: The data range is considered insufficient. Too small, aspect ratio If the value is too low or the proportion of outliers is too high, the distribution is considered irregular. When these conditions are met, PCA / robust fitting is preferred to improve stability. In specific implementation, when a preset condition is met, this set of parameters is recorded as valid calibration parameters. The preset condition is... This set of parameters includes at least: It is used for lane deviation calculation; it can also include fitting quality indicators (such as RMSE, R²) and applicable bay ranges (such as minBay, maxBay) for validity judgment and scope of application.
[0037] RMSE is the root mean square error between the model's predicted values and the fitted input values. The formula for calculation is as follows:
[0038] in, The predicted value output by the model. The input values used for fitting, The smaller the RMSE, the better the fit.
[0039] R² is the coefficient of determination, and its formula is as follows: in, The predicted value output by the model. The input values used for fitting, The mean, The sample size is denoted by R². The closer R² is to 1, the stronger the explanatory power of the model.
[0040] RMSE and R² are calculated on the training or validation set after fitting is completed, and are used to determine whether the parameter set is effective.
[0041] In practice, for trajectory points of multiple bays within the same yard, the average coordinates are calculated based on the bay number and projected onto the main axis of the driveway. A mapping between bay number and main axis coordinates is established through linear regression to estimate the spatial location of the yard boundary bay. The boundary bay is the bay with the smallest or largest yard bay number.
[0042] Specifically, the mapping between the B-value and the principal axis coordinates is established through linear regression as follows: Step 1: Calculate the representative point of the beta position By bay number Grouping, and calculating the average coordinates of the GNSS receiver's trajectory points for each beta, yields the representative beta point. :
[0043] Step 2: Project onto the lane main axis Based on the lane model parameters obtained from the fitting, the representative points of the lane position are projected onto the lane main axis.
[0044] First, calculate the distance:
[0045] Next, find the projection point:
[0046] Obtain the coordinates of the principal axis point corresponding to each bay. .
[0047] Step 3: Linear Regression Using the position sign as the independent variable, perform linear regression on the principal axis coordinates:
[0048] get Four parameters.
[0049] Step 4: Estimate the location of the boundary bay. Let the boundary position of the storage yard be... The spatial location of the boundary bay is estimated as follows:
[0050] In practice, if the spatial distance between two storage yards is less than the second threshold, they are judged to be physically adjacent storage yards. Based on the boundary points of the two storage yards and their respective lane models, a lane deviation model across storage yards is constructed for RTG deviation calculation in the cross-storage yard area.
[0051] Specifically, cross-yard areas can be determined by bay number exceeding the boundary or spatial proximity: when the bay number is less than 0 or greater than the maximum bay number of the current yard, it is determined to be a cross-yard state; by comparing the spatial distance of the bay numbers of adjacent yard boundaries, when the distance is less than a second threshold, it is determined to be a physically adjacent yard. Based on this, lane deviation model parameters for the cross-yard area can be constructed based on the boundary points of the two yards and their respective lane deviation model parameters. For example, a linear model can still be used, and this model can be called to calculate the lane deviation in the cross-yard state, thereby ensuring continuous deviation output during yard switching.
[0052] In practice, during normal RTG operation, the system receives raw data from the GNSS receiver in real time and looks up the calibration parameters of the current stockyard and corresponding equipment for each piece of raw data: if the position is within the legal range of the stockyard, the real-time lane deviation is calculated using the calibration model within the stockyard; if the position exceeds the legal range and there is a cross-stockyard correction model, the lane deviation model across the stockyard is called to calculate the deviation.
[0053] Specifically, when the position number satisfy The location of the shell is within the legal area of the storage yard, among which, Provided by the yard database or yard coordinate system. If the bit position number is outside the valid range, it is determined to be a cross-yard status, and the cross-yard correction model is invoked or a preset processing strategy is adopted.
[0054] In practice, the virtual lane deviation on the opposite side is generated using the following formula:
[0055] in, For the virtual lane deviation on the opposite side, For the actual side lane deviation, For the current gantry angle, This represents the current gantry length.
[0056] In practice, the linear parameters of the equipment and the cross-yard parameters fitted by the multiple sets of valid calibration samples for each yard are serialized into JSON format and saved in the local file system.
[0057] In summary, this application provides an automatic calibration and correction method for GNSS lane deviation of a port gantry crane. The method includes the following steps: acquiring raw data from a GNSS receiver, including gnssX, gnssY, angle, and status; acquiring lane line attitude data output by a camera, including camera lane deviation and detection status; acquiring the position and attitude of the RTG in the yard coordinate system, including the yard and bay position, and the attitude including pitch angle, roll angle, and gantry height; matching the raw data from the GNSS receiver and the lane line attitude data output by the camera within a preset time based on a timestamp; reading the yard and bay position, filtering data where the absolute value of the offset between the yard edge bay position and the center exceeds a first threshold, and obtaining multiple sets of valid calibration samples. Each set of data in the multiple sets of valid calibration samples includes gnssX, gnssY, laneDevFromCamera, ... Yard, bay, timestamp; Obtain the Euler angles and Euclidean transformations corresponding to the yard and bay position; Calculate the ground displacement deviation of the GNSS receiver in the trolley direction and gantry direction based on the gantry height; In the fitting stage, add the tilt compensation amount to the lane deviation value provided by the camera so that the fitting parameters reflect the geometric relationship under the tilt-free state; In the real-time deviation calculation stage, subtract the tilt compensation amount from the predicted deviation; For each yard and each GNSS receiver, collect calibration samples aggregated by bay position, and trigger fitting after reaching a preset value. Using GNSS coordinates and the tilt-compensated lane deviation as input, fit a linear model. By fitting the lane line information detected by the camera and the deviation relationship between the GNSS receiver and the lane centerline, automatic lane correction is achieved; Furthermore, the introduction of centralization The linear fitting model with constraints and tilt compensation restricts the mapping between the GNSS receiver and the lane centerline to physically reasonable rotation and translation, eliminating scaling errors. Furthermore, the calibration data can be automatically collected using the lane line detection results during normal RTG operation, without the need for additional point deployment and dedicated debugging time; Furthermore, it supports fully automatic timed calibration, batch calibration by stockpile, and manual trigger calibration, with multiple modes that can be flexibly combined; Furthermore, by combining attitude sensors, the influence of pitch / roll angle on the ground projection of the GNSS receiver is explicitly modeled, enabling stable output of high-precision lane deviation even under complex working conditions such as changes in spreader height, uneven ground, and off-center loading. Furthermore, even if a GNSS receiver is installed on only one side of the gantry or if one side of the GNSS receiver fails, the virtual lane deviation data on the other side can be calculated in real time based on the gantry length and attitude angle, providing the upper control system with input equivalent to dual-side GNSS. Furthermore, for adjacent storage yards within the same channel, a geometric interpolation method based on boundary position coordinates and their respective deviation models is used to construct a transition deviation model across storage yards; when the RTG moves from one storage yard to another, the lane deviation output is smooth and continuous, and there will be no sudden jump in error due to storage yard switching.
[0058] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims.
[0059] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.
Claims
1. A method for automatic calibration and correction of GNSS lane deviation for port gantry cranes, characterized in that, The method includes the following steps: Acquire raw data from the GNSS receiver, including gnssX, gnssY, angle, and status; Acquire lane line attitude data output by the camera, the lane line attitude data including camera lane deviation and detection status; Obtain the position and attitude of the RTG in the yard coordinate system, wherein the position includes the yard and bay position, and the attitude includes pitch angle, roll angle and gantry height; Based on the timestamp, the raw data from the GNSS receiver and the lane line attitude data output by the camera are matched within a preset time period; The data of the stockpile and bay position are read, and the data of the absolute value of the offset between the edge bay position and the center of the stockpile that exceeds the first threshold are filtered to obtain multiple sets of valid calibration samples. Each set of data in the multiple sets of valid calibration samples includes gnssX, gnssY, laneDevFromCamera, yard, bay, and timestamp. Obtain the Euler angles and Euclidean transformations corresponding to the stockpile and the position, and calculate the ground displacement deviation of the GNSS receiver in the trolley direction and the gantry direction based on the gantry height; During the fitting phase, the lane deviation value provided by the camera is added with the tilt compensation amount so that the fitting parameters reflect the geometric relationship under the tilt-free state. During the real-time deviation calculation phase, the tilt compensation amount is subtracted from the predicted deviation. For each storage yard and each GNSS receiver, calibration samples aggregated by position are collected. Once a preset value is reached, fitting is triggered. A linear model is fitted using GNSS coordinates and tilt-compensated lane deviation as inputs.
2. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 1, characterized in that, The tilt-compensated lane deviation is calculated using the following formula: Here, represents the lane deviation after tilt compensation. The geometric center of the sample coordinates is used for coordinate translation. For unit normal vector components, This is the intercept term.
3. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 1, characterized in that, Fit a linear model using the following method: When the data linearity is high and the noise is low, use a band... The least squares fitting method with constraints ensures that the normal vector is of unit length; When there are few data points or the distribution is irregular, a PCA-based principal axis / normal direction fitting method is adopted.
4. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 3, characterized in that, When a preset condition is met, this set of parameters will be recorded as valid calibration parameters. The preset condition is: .
5. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 1, characterized in that, For the trajectory points of multiple bays within the same yard, the average coordinates are calculated according to the bay and projected onto the main axis of the lane. A mapping between the bay number and the main axis coordinates is established through linear regression, which is used to estimate the spatial location of the bays at the yard boundary.
6. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 4, characterized in that, If the spatial distance between two storage yards is less than the second threshold, they are judged to be physically adjacent storage yards. Based on the boundary points of the two storage yards and their respective lane models, a lane deviation model across storage yards is constructed for RTG deviation calculation in the cross-storage yard area.
7. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 6, characterized in that, During normal RTG operation, the system receives raw data from the GNSS receiver in real time and looks up the calibration parameters of the current stockyard and corresponding equipment for each piece of raw data: if the position is within the legal range of the stockyard, the real-time lane deviation is calculated using the calibration model within the stockyard; if the position is outside the legal range and there is a cross-stockyard correction model, the lane deviation model across the stockyard is called to calculate the deviation.
8. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 1, characterized in that, The virtual lane deviation on the opposite side is generated using the following formula: in, For the virtual lane deviation on the opposite side, For the actual side lane deviation, For the current gantry angle, This represents the current gantry length.
9. The method for automatic calibration and correction of GNSS lane deviation for port gantry cranes according to claim 1, characterized in that, The linear parameters of the equipment and the cross-farm parameters fitted from the multiple sets of valid calibration samples for each farm are serialized into JSON format and saved in the local file system.