A roadway anchor rod identification and drilling positioning method based on multi-source fusion and geodesic extrapolation

Abstract

The present application relates to a kind of roadway anchor rod identification and drilling positioning method based on multi-source fusion and geodesic extrapolation, successively including the following steps: 1) based on the coordinate extraction of anchor rod that has been punched of double channel network;2) coordinate conversion based on error state kalman filter;3) the MLS fitting of roadway wall implicit manifold surface;4) based on the hole site extrapolation of four-order Runge-Kutta method;5) drilling machine attitude vector calculation.The present application is through the attention pooling of double mode feature and weight multiplication and addition calculation, when the distortion occurs in single mode sensor input data, algorithm maintains the accuracy of coordinate extraction process according to the adjustment of weight matrix.The present application uses RK4 algorithm to carry out geodesic numerical integration and extrapolate hole site, instead of three-dimensional vector straight line addition calculation, and the hole site spacing solving method is constrained on the geometry profile of rock surface;While solving the partial derivative of the coordinate point on the surface equation, provide normal attitude control input for drilling machine.

Description

Technical Field

[0001] This invention relates to a method for identifying roadway anchor bolts and locating boreholes based on multi-source fusion and geodesic extrapolation. Specifically, it relates to a method for extracting the spatial coordinates of anchor bolts under multi-modal sensors and for numerically calculating the new borehole pose under irregular rock surface conditions, belonging to the field of intelligent tunneling and support technology in coal mines. Background Technology

[0002] Rock bolt support in coal mine roadways is a core method for controlling surrounding rock deformation. In automated support operations, the spacing and installation angle of the rock bolts must comply with safety regulations. Before starting a new round of drilling, automated rock bolt drilling rigs must identify the actual spatial position of the previously installed rock bolts and use this as a benchmark to accurately deduce the three-dimensional target point and drilling posture of the new borehole. Existing technologies have the following shortcomings:

[0003] 1. The working face contains high concentrations of coal dust and water mist, which reduces the edge feature extraction rate of the visual extraction algorithm in the dusty environment; the curvature change of the lidar point cloud in the anchor tray area is not obvious, and the coordinate output of the recognition algorithm that relies on a single mode is unstable.

[0004] 2. The pose calculation during the movement of the drilling rig mostly relies on wheeled odometers or single-frame point cloud registration. Due to the lack of high-frequency motion compensation, the calculation results have cumulative errors.

[0005] 3. Hole location calculations often employ linear translations of three-dimensional spatial coordinates, failing to consider the undulating characteristics of the tunnel wall. On uneven rock surfaces, the spacing calculated by linear translations deviates physically from the actual arc length against the wall; furthermore, this calculation method cannot output the normal vector of the local rock surface, potentially leading to eccentric pressure when the drilling rig advances at a preset fixed angle. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method for identifying roadway anchor bolts and locating boreholes based on multi-source fusion and geodesic extrapolation. This method is based on multimodal data processing and numerical calculation to obtain the spatial coordinates of the anchor bolts and borehole target point parameters that conform to the wall arc length constraint.

[0007] The technical solution of the present invention is as follows:

[0008] A method for identifying roadway anchor bolts and locating boreholes based on multi-source fusion and geodesic extrapolation includes the following steps:

[0009] 1) Coordinate extraction of installed anchor bolts based on dual-channel network

[0010] Using machine vision to synchronously acquire RGB images of the tunnel walls With 3D point cloud ;Will Inputting a ResNet backbone network yields two-dimensional texture feature maps. ;Will Input PointNet++ to obtain 3D geometric feature maps ;

[0011] The spatially registered 2D texture feature map With three-dimensional geometric feature map The data is concatenated and subjected to global average pooling and global max pooling. The modal adaptive weight matrix is ​​then calculated using a multilayer perceptron. The modal adaptive weight matrix is ​​used to characterize the contribution of RGB texture information and point cloud geometric information under the current working conditions.

[0012]

[0013] Computational fusion features fusion features Based on the modal adaptive weight matrix It is obtained by weighted fusion of two-dimensional texture features and three-dimensional geometric features, i.e. The network output layer is based on Return, output the coordinates of the anchor bolts in the sensor coordinate system. ;

[0014] 2) Coordinate transformation based on error state Kalman filtering

[0015] Define the state vector of the drilling rig chassis. ;in, This represents the position of the drilling rig chassis in the global coordinate system. For speed, For attitude quaternions, To achieve zero bias in the gyroscope, To achieve zero bias in the accelerometer, For gravitational acceleration; the error state vector is given by , , , , Composition, used for ESKF prediction and observation updates;

[0016] Collect the angular velocity and acceleration of the inertial measurement unit and substitute them into the kinematic equations:

[0017] , , , , Pre-integration is performed on the state vector and the error state vector.

[0018] in, The measured angular velocity of the gyroscope. Accelerometer measured specific force Gaussian white noise for gyroscope angular velocity. Add specific Gaussian white noise. , Zero-biased random walk noise;

[0019] Two consecutive frames of LiDAR point clouds are extracted, and the relative pose matrix, i.e., the relative pose matrix from the LiDAR coordinate system of frame k to the LiDAR coordinate system of frame (k+1), is calculated using the iterative nearest-point algorithm. Then, based on the calibration extrinsic parameters of the LiDAR relative to the drilling rig chassis coordinate system, the inter-frame relative pose of the LiDAR is converted into inter-frame relative pose observations of the drilling rig chassis, which are used as ESKF measurement update inputs to correct the state vector of the drilling rig chassis. ;

[0020] The ESKF state estimation module outputs the calculated global rotation matrix of the drilling rig. With translation vector By combining the calibration extrinsic parameters of the RGB camera, LiDAR, and IMU relative to the coordinate system of the drilling rig chassis, the global coordinates of the anchor bolt are calculated. :

[0021]

[0022] in, and These represent the external parameter rotation matrix and translation vector from the sensor coordinate system to the drilling rig chassis coordinate system, respectively.

[0023] 3) MLS fitting of implicit manifold surfaces on tunnel walls

[0024] Selecting coordinates including the reference anchor bolt and point cloud collection on the rock surface in front. The point cloud set is the LiDAR local point cloud transformed to the global coordinate system after ESKF pose compensation in step 2). After removing outliers, the implicit manifold surface is constructed using the moving least squares method. ;

[0025] Define the weighted sum of squared orthogonal distances of the local fitting planes within the neighborhood of the sampling points on the rock surface to be fitted:

[0026]

[0027] in, A collection of point clouds on a local rock surface The Middle One sampling point, For the current query point, To fit the plane normal, The distance from the origin, the weight function Gaussian decay function , It is a smoothing factor;

[0028] minimize The curvature and continuous partial derivative equations at each point are obtained, providing a geometric basis for subsequent extrapolation of hole position and solution of drilling attitude under surface constraints.

[0029] 4) Hole position extrapolation based on the fourth-order Runge-Kutta method

[0030] Set the row spacing baseline value as The design axis vector of the tunnel is The coordinates of the outermost anchor bolt are used as the initial point. ;

[0031] On the implicit manifold surface, the tunnel design axis vector is projected onto the tangent plane at the initial point and normalized to obtain the initial tangent vector; the geodesic integration direction is defined by the surface normal constraint and the unit arc length constraint, so that the extrapolated trajectory always lies on the tunnel wall surface:

[0032]

[0033] Construct with arc length Differential equations with integration variables Set the integration step size to be The fourth-order Runge-Kutta method is used for iteration: at each step, the tangent vector is updated based on the local surface normal, and the arc length attached to the wall is accumulated.

[0034]

[0035]

[0036]

[0037]

[0038]

[0039] When the cumulative steps The loop terminates when the time is right, and the final result is obtained. Let the target coordinates be denoted as . That is, the termination condition is that the cumulative arc length against the wall reaches the row spacing benchmark value;

[0040] 5) Calculation of drilling rig attitude vector

[0041] Based on surface functions ,calculate The three-dimensional partial derivatives at a given point are normalized to obtain the attitude vector. When the design drilling direction requires perpendicularity to the rock surface, the unit normal vector is used as the drill rod attitude vector; when the design drilling direction has a preset inclination angle, the attitude is corrected based on the unit normal vector according to the designed inclination angle.

[0042]

[0043] System output control command parameter matrix The drilling rig will then perform the corresponding actions.

[0044] Preferably, in step 1), the RGB image and the three-dimensional point cloud are acquired by an explosion-proof industrial camera and a solid-state LiDAR that have completed time synchronization and extrinsic parameter calibration; based on the camera intrinsic parameters and the camera-LiDAR extrinsic parameters, the LiDAR point cloud is projected onto the RGB image coordinate system to establish a spatial correspondence between the image pixel region and the three-dimensional point neighborhood.

[0045] Preferably, in step 4), the row spacing reference value is set to... Set the integration step size to be .

[0046] A computer device includes an input / output unit, a memory, and a processor, wherein the memory stores computer-readable instructions that, when executed by the processor, cause the processor to perform the steps in the method described above.

[0047] A storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps in the method described above.

[0048] Technical features and beneficial effects of the present invention:

[0049] 1. This invention uses attention pooling and weight multiplication and addition calculation of dual-modal features to maintain the accuracy of the coordinate extraction process when the input data of a single-modal sensor is distorted. The algorithm adjusts the weight matrix to achieve this.

[0050] 2. In this invention, the ESKF fusion framework uses high-frequency IMU data to compensate for the motion deformation of LiDAR data during the inter-frame registration process, thereby reducing the cumulative drift when mapping the drill rig coordinate system to the global coordinate system.

[0051] 3. This invention uses the RK4 algorithm to extrapolate borehole locations using geodesic numerical integration, replacing the three-dimensional vector linear addition calculation, and constrains the borehole spacing calculation method to the geometric contour of the rock surface; at the same time, it solves the partial derivative of the coordinate point on the surface equation, providing normal attitude control input for the drilling rig.

[0052] 4. By using the anchor bolt spatial coordinate extraction method under the multimodal sensor and the new borehole pose numerical calculation method under irregular rock surface conditions of the present invention, the spatial coordinates of the anchor bolt and the borehole target point parameters that conform to the wall arc length constraint can be accurately obtained. Attached Figure Description

[0053] Figure 1 This is a system architecture diagram of multi-source data processing and pose calculation in this invention;

[0054] Figure 2 This is a schematic diagram of the dual-channel neural network feature fusion structure in this invention. Detailed Implementation

[0055] The present invention will be further described below with reference to the embodiments and accompanying drawings, but is not limited thereto.

[0056] Terminology Explanation:

[0057] An RGB image is an image composed of three color channels: red, green, and blue. Each pixel contains the color information of that point, namely the (R, G, B) value.

[0058] A 3D point cloud is a data structure that represents an object or scene in 3D space. It is a set of 3D coordinate points (X, Y, Z) obtained by discrete sampling of the surface of an object / scene. It can also be supplemented with attributes such as color, reflection intensity, and timestamp.

[0059] The ResNet backbone network, or Residual Neural Network, is a convolutional neural network structure that addresses the gradient vanishing and accuracy degradation problems in deep networks by introducing the concept of residual learning.

[0060] PointNet++ is a deep learning framework specifically designed to process point cloud data. It is an improved version of PointNet that gradually increases the receptive field through a hierarchical structure to capture geometric features at different scales.

[0061] LiDAR is an abbreviation for "Light Detection and Ranging," a remote sensing technology that uses rapidly emitted laser pulses to map the Earth's surface and collect and map detailed information and accurate 3D models of objects.

[0062] ESKF (Error State Kalman Filter) is a filtering method specifically designed for inertial navigation. Its core is to estimate the error of the nominal state rather than the complete state itself, thereby more naturally modeling the error characteristics of the IMU (Inertial Measurement Unit) (such as zero bias and scale factor), and improving numerical stability and engineering robustness.

[0063] Moving Least Squares (MLS) is a non-parametric numerical analysis technique used for constructing and interpolating smooth data surfaces.

[0064] Computational fusion features refer to integrating feature information from different sources, levels, or modalities to obtain a more comprehensive and discriminative feature representation, thereby improving model performance.

[0065] See Figure 1 , Figure 2 This invention provides a method for identifying roadway anchor bolts and locating boreholes based on multi-source fusion and geodesic extrapolation, specifically including the following steps:

[0066] 1) Coordinate extraction of installed anchor bolts based on dual-channel network

[0067] Using machine vision to synchronously acquire RGB images of the tunnel walls With 3D point cloud ;Will Input ResNet backbone network to extract 2D texture feature maps ;Will Input PointNet++ to obtain 3D geometric feature maps The RGB image and the 3D point cloud are acquired by an explosion-proof industrial camera and a solid-state LiDAR that have completed time synchronization and extrinsic parameter calibration. Based on the camera intrinsic parameters and camera-LiDAR extrinsic parameters, the LiDAR point cloud is projected onto the RGB image coordinate system to establish the spatial correspondence between the image pixel region and the 3D point neighborhood.

[0068] After concatenating the above feature maps, global average pooling (AvgPool) and global max pooling (MaxPool) are performed. The modality adaptive weight matrix is ​​then calculated using a multilayer perceptron (MLP). The modal adaptive weight matrix is ​​used to characterize the contribution of RGB texture information and point cloud geometric information under the current working conditions.

[0069]

[0070] Computational fusion features fusion features The two-dimensional texture features and three-dimensional geometric features are obtained by weighted fusion based on the modal adaptive weight matrix. The network output layer is based on Return, output the coordinates of the anchor bolts in the sensor coordinate system. ;

[0071] 2) Coordinate transformation based on error state Kalman filter (ESKF)

[0072] Define the state vector of the drilling rig chassis. ;in, This represents the position of the drilling rig chassis in the global coordinate system. For speed, For attitude quaternions, To achieve zero bias in the gyroscope, To achieve zero bias in the accelerometer, For gravitational acceleration; the error state vector is given by , , , , It is used for ESKF prediction and observation updates.

[0073] Collect the angular velocity and acceleration of the inertial measurement unit (IMU) and substitute them into the kinematic equations:

[0074] , , , , Pre-integration is performed on the state vector and the error state vector.

[0075] in, The measured angular velocity of the gyroscope. Accelerometer measured specific force Gaussian white noise for gyroscope angular velocity. Add specific Gaussian white noise. , Zero-biased random walk noise;

[0076] Two consecutive frames of LiDAR point clouds are extracted, and the relative pose matrix, i.e., the relative pose matrix from the LiDAR coordinate system of frame k to the LiDAR coordinate system of frame (k+1), is calculated using the iterative nearest-point algorithm. Then, based on the calibration extrinsic parameters of the LiDAR relative to the drilling rig chassis coordinate system, the inter-frame relative pose of the LiDAR is converted into inter-frame relative pose observations of the drilling rig chassis, which are used as ESKF measurement update inputs to correct the state vector of the drilling rig chassis. ;

[0077] The ESKF state estimation module outputs the calculated global rotation matrix of the drilling rig. With translation vector By combining the calibration extrinsic parameters of the RGB camera, LiDAR, and IMU relative to the coordinate system of the drilling rig chassis, the global coordinates of the anchor bolt are calculated. :

[0078]

[0079] in, and These represent the external parameter rotation matrix and translation vector from the sensor coordinate system to the drilling rig chassis coordinate system, respectively.

[0080] 3) MLS fitting of implicit manifold surfaces on tunnel walls

[0081] Selecting coordinates including the reference anchor bolt and point cloud collection on the rock surface in front. The point cloud set is the LiDAR local point cloud transformed to the global coordinate system after ESKF pose compensation in step 2). After removing outliers, the implicit manifold surface is constructed using the moving least squares (MLS) method. ;

[0082] Define the weighted sum of squared orthogonal distances of the local fitting planes within the neighborhood of the sampling points on the rock surface to be fitted:

[0083]

[0084] in, A collection of point clouds on a local rock surface The Middle One sampling point, For the current query point, To fit the plane normal, The distance from the origin, the weight function Gaussian decay function , It is a smoothing factor;

[0085] minimize The curvature and continuous partial derivative equations at each point are obtained, providing a geometric basis for subsequent extrapolation of hole position and solution of drilling attitude under surface constraints.

[0086] 4) Hole position extrapolation based on the fourth-order Runge-Kutta method (RK4)

[0087] Set the row spacing baseline value as The design axis vector of the tunnel is The coordinates of the outermost anchor bolt are used as the initial point. Among them, the tunnel design axis vector serves as the global reference direction for extrapolating the borehole position, and the initial point is the global coordinate of the outermost identified anchor bolt along the extrapolation direction.

[0088] On the implicit manifold surface, the tunnel design axis vector is projected onto the tangent plane at the initial point and normalized to obtain the initial tangent vector; the geodesic integration direction is defined by the surface normal constraint and the unit arc length constraint, ensuring that the extrapolated trajectory always lies on the tunnel wall surface.

[0089]

[0090] Construct with arc length Differential equations with integration variables Set the integration step size to Industrial control computer The point cloud of the region is obtained by performing MLS fitting. The tangent vector constraint equation is invoked to perform iterative integration based on the RK4 algorithm until the accumulated arc length meets the row spacing requirement, and the coordinates of the borehole target point are output. That is, the fourth-order Runge-Kutta method (RK4) is used for iteration: at each step, the tangent vector is updated according to the local surface normal, and the arc length attached to the wall is accumulated.

[0091]

[0092]

[0093]

[0094]

[0095]

[0096] When the accumulated steps The loop terminates when the time is right, and the final result is obtained. Let the target coordinates be denoted as . That is, the termination condition is that the cumulative arc length against the wall reaches the row spacing benchmark value.

[0097] 5) Calculation of drilling rig attitude vector

[0098] Based on surface functions ,calculate The three-dimensional partial derivatives at a given point are normalized to obtain the attitude vector. Wherein, when the design drilling direction requires perpendicularity to the rock surface, the unit normal vector is used as the drill rod attitude vector; when the design drilling direction has a preset inclination angle, the attitude is corrected according to the design inclination angle based on the unit normal vector.

[0099]

[0100] System output control command parameter matrix The drive robotic arm adjusts the drill pipe axis to Orientation, drill bit tip aligned To carry out drilling operations.

[0101] To implement the embodiments, the present invention also proposes an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps in the aforementioned method.

[0102] To implement the embodiments, the present invention also proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the tunnel anchor bolt identification and borehole positioning method based on multi-source fusion and geodesic extrapolation as described in the embodiments of the present invention.

[0103] In this embodiment, the functional units can be integrated into one processing module, or each unit can exist physically separately, or two or more units can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium mentioned can be a read-only memory, a hard disk, or an optical disk, etc.

[0104] The above description is merely a specific embodiment of the present invention, and the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying roadway anchor bolts and locating boreholes based on multi-source fusion and geodesic extrapolation, characterized in that, Includes the following steps: 1) Coordinate extraction of installed anchor bolts based on dual-channel network Using machine vision to synchronously acquire RGB images of the tunnel walls With 3D point cloud ;Will Inputting a ResNet backbone network yields two-dimensional texture feature maps. ;Will Input PointNet++ to obtain 3D geometric feature maps ; The spatially registered 2D texture feature map With three-dimensional geometric feature map The data is concatenated and subjected to global average pooling and global max pooling. The modal adaptive weight matrix is ​​then calculated using a multilayer perceptron. The modal adaptive weight matrix is ​​used to characterize the contribution of RGB texture information and point cloud geometric information under the current working conditions. Computational fusion features fusion features Based on the modal adaptive weight matrix It is obtained by weighted fusion of two-dimensional texture features and three-dimensional geometric features, i.e. The network output layer is based on Return, output the coordinates of the anchor bolts in the sensor coordinate system. ; 2) Coordinate transformation based on error state Kalman filtering Define the state vector of the drilling rig chassis. ;in, This represents the position of the drilling rig chassis in the global coordinate system. For speed, For attitude quaternions, To achieve zero bias in the gyroscope, To achieve zero bias in the accelerometer, For gravitational acceleration; the error state vector is given by , , , , Composition, used for ESKF prediction and observation updates; Collect the angular velocity and acceleration of the inertial measurement unit and substitute them into the kinematic equations: , , , , Pre-integration is performed on the state vector and the error state vector. in, The measured angular velocity of the gyroscope. Accelerometer measured specific force Gaussian white noise for gyroscope angular velocity. Add specific Gaussian white noise. , Zero-biased random walk noise; Two consecutive frames of LiDAR point clouds are extracted, and the relative pose matrix, i.e., the relative pose matrix from the LiDAR coordinate system of frame k to the LiDAR coordinate system of frame (k+1), is calculated using the iterative nearest-point algorithm. Then, based on the calibration extrinsic parameters of the LiDAR relative to the drilling rig chassis coordinate system, the inter-frame relative pose of the LiDAR is converted into inter-frame relative pose observations of the drilling rig chassis, which are used as ESKF measurement update inputs to correct the state vector of the drilling rig chassis. ; The ESKF state estimation module outputs the calculated global rotation matrix of the drilling rig. With translation vector By combining the calibration extrinsic parameters of the RGB camera, LiDAR, and IMU relative to the coordinate system of the drilling rig chassis, the global coordinates of the anchor bolt are calculated. : in, and These represent the external parameter rotation matrix and translation vector from the sensor coordinate system to the drilling rig chassis coordinate system, respectively. 3) MLS fitting of implicit manifold surfaces on tunnel walls Selecting coordinates including the reference anchor bolt and point cloud collection on the rock surface in front. The point cloud set is the LiDAR local point cloud transformed to the global coordinate system after ESKF pose compensation in step 2). After removing outliers, the implicit manifold surface is constructed using the moving least squares method. ; Define the weighted sum of squared orthogonal distances of the local fitting planes within the neighborhood of the sampling points on the rock surface to be fitted: in, A collection of point clouds on a local rock surface The Middle One sampling point, For the current query point, To fit the plane normal, The distance from the origin, the weight function Gaussian decay function , It is a smoothing factor; minimize The curvature and continuous partial derivative equations at each point are obtained, providing a geometric basis for subsequent extrapolation of hole position and solution of drilling attitude under surface constraints. 4) Hole position extrapolation based on the fourth-order Runge-Kutta method Set the row spacing baseline value as The design axis vector of the tunnel is The coordinates of the outermost anchor bolt are used as the initial point. ; On the implicit manifold surface, the tunnel design axis vector is projected onto the tangent plane at the initial point and normalized to obtain the initial tangent vector; the geodesic integration direction is defined by the surface normal constraint and the unit arc length constraint, so that the extrapolated trajectory always lies on the tunnel wall surface: Construct with arc length Differential equations with integration variables Set the integration step size to The fourth-order Runge-Kutta method is used for iteration: at each step, the tangent vector is updated based on the local surface normal, and the arc length attached to the wall is accumulated. When the accumulated steps The loop terminates when the time is right, and the final result is obtained. Let the target coordinates be denoted as . That is, the termination condition is that the cumulative arc length against the wall reaches the row spacing benchmark value; 5) Calculation of drilling rig attitude vector Based on surface functions ,calculate The three-dimensional partial derivatives at a given point are normalized to obtain the attitude vector. When the design drilling direction requires perpendicularity to the rock surface, the unit normal vector is used as the drill rod attitude vector; when the design drilling direction has a preset inclination angle, the attitude is corrected based on the unit normal vector according to the designed inclination angle. System output control command parameter matrix The drilling rig will then perform the corresponding actions.

2. The method for identifying roadway anchor bolts and locating boreholes based on multi-source fusion and geodesic extrapolation as described in claim 1, characterized in that, In step 1), the RGB image and the 3D point cloud are acquired by an explosion-proof industrial camera and a solid-state LiDAR that have completed time synchronization and extrinsic parameter calibration; based on the camera intrinsic parameters and the camera-LiDAR extrinsic parameters, the LiDAR point cloud is projected onto the RGB image coordinate system to establish the spatial correspondence between the image pixel region and the 3D point neighborhood.

3. The method for identifying roadway anchor bolts and locating boreholes based on multi-source fusion and geodesic extrapolation as described in claim 1, characterized in that, In step 4), the row spacing reference value is set to... Set the integration step size to .

4. A computer device comprising an input / output unit, a memory, and a processor, wherein the memory stores computer-readable instructions, which, when executed by the processor, cause the processor to perform the steps of any of the methods of claims 1-3.

5. A storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of any of the methods of claims 1-3.

Images