Irregular roadway wall surface fitting method

Abstract

The application discloses an irregular roadway wall surface fitting method, which comprises the following steps: firstly, collecting laser pixel planes offline: static measurement of the data of a dug roadway wall surface, point cloud splicing, generation of an effective laser data matrix, establishment of a laser fitting plane coordinate system, generation of an effective laser pixel plane, maximum error calculation and saving; secondly, real-time measurement of a roadway wall surface normal vector and distance: loading of the effective laser pixel plane, generation of effective ultrasonic array data, fitting of a space plane under the ultrasonic array coordinate system, average error calculation, when the average error is smaller than the maximum error, calculation and saving of the normal vector; otherwise, Gaussian sampling is carried out, a new fitting plane is generated, and the distance from the ultrasonic array center to the plane is calculated. The application can realize accurate estimation of the normal vector of the irregular roadway wall surface and the distance from the measuring device to the plane, can provide relative position and attitude information in the roadway section for a heading machine, has good use effect, and is convenient to popularize and use.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of plane fitting, and particularly relates to an irregular roadway wall surface fitting method. BACKGROUND

[0002] In the process of underground coal mining, the heading machine usually calculates the distance relative to the two sides of the roadway wall to calculate the degree of deviation from the center line, so as to avoid lateral deviation during operation. In the traditional method, single ranging sensor or multi-ranging sensor arbitration is generally used to complete ranging, but due to the unevenness of the plane excavated by the heading machine, the distance error calculated by this method is large. If a measuring and fitting method is used to consider the case of uneven plane, the plane of the roadway excavated by the heading machine is fitted, and then the distance from the measuring device to the fitted plane of the roadway wall is accurately calculated, the relative position of the measuring device in the roadway cross section can be obtained, which plays an important role in the relative positioning of the heading machine to the roadway wall. In the actual environment, the roadway wall excavated by the heading machine is an irregular plane, and in the actual operation of the heading machine, the roadway is full of dust, so it is a key technology to design a high-precision plane fitting method in such a complex environment.

[0003] In the prior art, there is a lack of a fitting method for irregular roadway wall surfaces in the real-time heading process of a tunnel or underground mine heading machine, which obtains the plane normal and the distance from the measuring device to the fitted plane, and has high precision and robustness. SUMMARY

[0004] The technical problem to be solved by the present application is to provide an irregular roadway wall surface fitting method, which has the advantages of simple method steps, reasonable design, easy implementation, fitting of irregular roadway wall surfaces based on an ultrasonic array, accurate estimation of the normal vector of the irregular roadway wall surface and the distance from the measuring device to the plane during the mine roadway heading process, small distance estimation error and normal vector angle error, provision of relative position and attitude information of the roadway cross section for the heading machine, good use effect, and easy popularization and use.

[0005] To solve the above technical problems, the technical scheme adopted by the present application is as follows: an irregular roadway wall surface fitting method, comprising the following steps:

[0006] Step one, offline collection of a laser pixel plane;

[0007] Step 101, static measurement of the excavated roadway wall surface data by using a scanner device composed of a 3D laser radar and a high-precision inertial navigation system, saving of 3D laser radar data and pose data;

[0008] Step 102, point cloud splicing according to the 3D laser radar data and pose data in the initial frame space coordinate system;

[0009] Step 103, intercepting point cloud in a set range to generate an effective laser data matrix;

[0010] Step 104, fitting a plane model initial value according to the effective laser data matrix, and solving an optimal plane model by an optimization method to establish a laser fitted plane coordinate system;

[0011] Step 105, projecting the first frame optimal space plane to the plane coordinate system under the plane to generate an effective laser pixel plane;

[0012] Step 106, taking the sum of absolute values of pixel values corresponding to the ultrasonic array as an error, traversing the effective laser pixel plane to find the maximum error and save it;

[0013] Step two, real-time measurement of roadway wall normal vector and distance;

[0014] Step 201, loading the effective laser pixel plane;

[0015] Step 202, real-time acquisition of ultrasonic array data, and using RANSAC algorithm to remove outliers to generate effective ultrasonic array data;

[0016] Step 203, when the number of inliers is greater than or equal to 10, the next step is executed; when the number of inliers is less than 10, the position of the ultrasonic array is adjusted, and steps 202-203 are re-executed;

[0017] Step 204, fitting a space plane in the ultrasonic array coordinate system using the LM algorithm;

[0018] Step 205, calculating the pose relationship between the real-time fitted plane and the effective laser pixel plane, and projecting the space plane to the space plane coordinate system;

[0019] Step 206, calculating the error between the real-time acquisition value and the corresponding effective laser pixel plane value, traversing the effective laser pixel plane with a set value P in the horizontal and vertical directions, repeating step 205, recording the error each time, and calculating the average error;

[0020] Step 207, when the average error is less than the maximum error in step 106, the normal vector of the space plane in step 204 is calculated and saved; when the average error is greater than or equal to the maximum error in step 106, the space plane in step 204 is sampled by Gauss to generate a new fitted plane and calculate the distance from the center of the ultrasonic array to the plane.

[0021] The above-mentioned irregular roadway wall fitting method, the specific process of fitting the plane model initial value in step 104 includes:

[0022] Step A1, establishing a plane model;

[0023] A L x+B L y+C L z+D L =0

[0024] Wherein, (A L ,B L ,C L ,D L ) is the plane model parameter to be solved, (x, y, z) is the point variable of the plane;

[0025] Step A2, in the point cloud in the set range, three points are randomly substituted into the model, and the model parameters are calculated;

[0026] Step A3, traverse the remaining points in the point cloud in the set range, and calculate the distance from each point to the plane;

[0027] Step A4, set the points with a point-plane distance less than a numerical value M as inner points, and the points with a point-plane distance greater than or equal to the numerical value M as outer points;

[0028] Step A5, update the plane model;

[0029] Step A6, repeat steps A1-A4, when the number of inner points is greater than the number of inner points in the last cycle, continue to update the plane model; Otherwise, save the model parameters.

[0030] The specific process of the optimization method in step 104 of the above irregular roadway wall surface fitting method includes:

[0031] Step B1, establish an optimization model;

[0032]

[0033] Wherein, (A, B, C, D) is the parameter to be optimized, (x i ,y i ,z i ) is the 3D coordinates of the i-th point in the point cloud, and N is the number of points included in the point cloud;

[0034] Step B2, use the gradient descent method to solve the optimal plane model parameters (A * ,B * ,C * ,D * ) with the model parameters in step A6 as the initial value.

[0035] The pixel number of the effective laser pixel plane in step 105 of the above irregular roadway wall surface fitting method is 2000*100, and the pixel horizontal and vertical spacing corresponds to an actual distance of 1cm. Each pixel includes a vertical coordinate in the laser fitting plane coordinate system.

[0036] The irregular roadway wall surface fitting method, wherein the set value P in step 206 is 5 cm.

[0037] The irregular roadway wall surface fitting method, wherein the specific process of the Gaussian sampling in step 207 comprises:

[0038] Step C1, random numbers (μ1, μ2) are taken in the range of [0, 1], and random parameters (ΔA, ΔB) are taken:

[0039]

[0040] Step C2, new plane parameters are calculated according to (ΔA, ΔB):

[0041]

[0042] Step C3, the new plane parameters are fitted as a space plane.

[0043] The irregular roadway wall surface fitting method, wherein the ultrasonic wave array is arranged in a 100 cm*50 cm plane, 12 ultrasonic wave sensors are uniformly arranged, 4 in the horizontal direction and 3 in the vertical direction, and the horizontal and vertical spacing between the sensors is 25 cm.

[0044] Compared with the prior art, the method has the following advantages: the method steps are simple, the design is reasonable, the implementation is convenient, the irregular roadway wall surface is fitted based on the ultrasonic wave array, the normal vector of the irregular roadway wall surface and the distance from the measuring device to the plane can be accurately estimated during the mine roadway tunneling process, the distance estimation error and the normal vector angle error are small, the relative position and attitude information in the roadway cross section can be provided for the tunneling machine, the use effect is good, and the method is convenient to popularize and use.

[0045] The technical solutions of the present application will be further described in detail below with reference to the drawings and embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 The method flowchart of the present application;

[0047] Figure 2 The method flowchart of the present application for offline collection of laser pixel planes;

[0048] Figure 3 The method flowchart of the present application for real-time measurement of roadway wall surface normal vectors and distances;

[0049] Figure 4 The schematic diagram of the effective laser pixel plane of the present application;

[0050] ​Figure 5 A coordinate conversion diagram of the present application. DETAILED DESCRIPTION

[0051] As Figures 1-3 shown, the irregular roadway wall surface fitting method of the present application comprises the following steps:

[0052] Step one, offline collection of laser pixel plane;

[0053] Step 101, static measurement of the excavated roadway wall surface data using a scanner device composed of a 3D laser radar and a high-precision inertial navigation system, saving 3D laser radar data and pose data;

[0054] Step 102, point cloud stitching according to the 3D laser radar data and pose data in the initial frame space coordinate system;

[0055] Step 103, intercepting point clouds within a set range to generate an effective laser data matrix;

[0056] In specific implementation, the set range is 20m*1m.

[0057] Step 104, fitting a plane model initial value according to the effective laser data matrix, and solving the optimal plane model using an optimization method to establish a laser fitted plane coordinate system;

[0058] The specific process of fitting the plane model initial value includes:

[0059] Step A1, establishing a plane model;

[0060] A L x+B L y+C L z+D L =0

[0061] Wherein, (A L ,B L ,C L ,D L ) are the plane model parameters to be solved, and (x,y,z) are the plane point variables;

[0062] Step A2, substituting three points in the point cloud within the set range into the model to calculate the model parameters;

[0063] Step A3, traversing the remaining points in the point cloud within the set range and calculating the distance from each point to the plane;

[0064] Step A4, setting points with a point-plane distance less than a numerical value M as inliers, and points with a point-plane distance greater than or equal to the numerical value M as outliers;

[0065] Step A5, updating the plane model;

[0066] Step A6: Repeat steps A1 to A4. If the number of interior points is greater than the number of interior points in the previous loop, continue to update the planar model; otherwise, save the model parameters.

[0067] The specific process of the optimization method includes:

[0068] Step B1: Establish an optimization model;

[0069]

[0070] Where (A,B,C,D) are the parameters to be optimized, (x i ,y i ,z i Let be the 3D coordinates of the i-th point in the point cloud, and N be the number of points in the point cloud;

[0071] Step B2: Using the gradient descent method, with the model parameters from step A6 as initial values, solve for the optimal planar model parameters (A). * B * C * D * ).

[0072] Step 105: Project the optimal spatial plane of the first frame onto the planar coordinate system below the plane to generate an effective laser pixel plane;

[0073] The effective laser pixel plane has 2000*100 pixels, the horizontal and vertical spacing of the pixels corresponds to an actual distance of 1cm, and each pixel includes the vertical coordinate in the laser fitting plane coordinate system.

[0074] In specific implementation, such as Figure 4 As shown, in O L -x L y L In the plane represented by the coordinate system, each pixel represents the distance from the measured value to the fitting plane. The number of pixels is 2000*100, and the horizontal and vertical spacing of the pixels corresponds to the actual distance of 1cm. Each pixel contains the vertical coordinate under the laser fitting plane coordinate system.

[0075] Step 106: Take the sum of the absolute values ​​of the corresponding pixel values ​​of the ultrasonic array as the error, traverse the effective laser pixel plane, find the maximum error and save it;

[0076] Step 2: Measure the normal vector and distance of the tunnel wall in real time;

[0077] Step 201: Load the effective laser pixel plane;

[0078] Step 202: Acquire ultrasonic array data in real time, use the RANSAC algorithm to remove outliers, and generate valid ultrasonic array data;

[0079] Step 203: When the number of inner points is greater than or equal to 10, proceed to the next step; when the number of inner points is less than 10, adjust the position of the ultrasonic array and repeat steps 202 to 203.

[0080] Step 204: Fit the spatial plane using the LM algorithm in the ultrasonic array coordinate system;

[0081] In specific implementation, such as Figure 5 As shown, in the ultrasonic array coordinate system O Tp -x Tp y Tp z Tp Next, the LM algorithm is used to fit the spatial plane (A) of the ultrasonic array data acquired in step 202. Tp B Tp C Tp D Tp The LM algorithm steps are as described in step 104, which is the optimization method step.

[0082] Further, the various ultrasonic measurement values ​​in the spatial coordinate system of the ultrasonic pixel plane are obtained, such as... Figure 5 As shown o p -x p y p z p Coordinate system; thus, the ultrasonic pixel plane coordinate system is obtained, such as Figure 5 As shown in O l -x l y l The coordinate system stores the error between the actual measurement value and the fitted plane for each pixel.

[0083] Step 205: Calculate the pose relationship between the real-time fitting plane and the effective laser pixel plane, and project the spatial plane onto the spatial plane coordinate system;

[0084] Step 206: Calculate the error between the real-time acquired value and the corresponding effective laser pixel plane value. Traverse the effective laser pixel plane with a horizontal and vertical interval set value P, repeat step 205, record the error each time, and calculate the average error.

[0085] The set value P is 5cm.

[0086] In practice, traversing O l -x l y l 12 ultrasonic measurement values ​​in the coordinate system (x i ,y i Its corresponding vertical coordinate is z. i According to O L -x L y L Four values ​​in the coordinate system The vertical coordinate in the corresponding effective laser plane coordinate system is Let:

[0087]

[0088]

[0089] The difference between the vertical coordinate of the ultrasonic measurement plane and the vertical coordinate of the effective laser pixel plane is calculated as

[0090] Step 207, when the average error is less than the maximum error in step 106, the normal vector of the spatial plane in step 204 is calculated and saved; when the average error is greater than or equal to the maximum error in step 106, the spatial plane in step 204 is Gauss sampled, a new fitting plane is generated and the distance from the center of the ultrasonic array to the plane is calculated;

[0091] The specific process of the Gauss sampling includes:

[0092] Step C1, random numbers (μ1, μ2) are taken in the range of [0, 1], and random parameters (ΔA, ΔB) are taken:

[0093]

[0094] Step C2, the new plane parameters are calculated according to (ΔA, ΔB):

[0095]

[0096] Step C3, the new plane parameters are taken as the new plane parameters fitting the spatial plane.

[0097] In the embodiment, the ultrasonic array is arranged in a 100cm*50cm plane, 12 ultrasonic sensors are uniformly arranged, 4 in the horizontal direction and 3 in the vertical direction, and the horizontal and vertical spacing between the sensors is 25cm.

[0098] The present application is based on the irregular plane fitting method of multiple ultrasonic arrays, the distance information is measured by the ultrasonic array, and the plane normal and the accurate distance from the measuring device are fitted.

[0099] The above is only a preferred embodiment of the present application, and does not limit the present application in any way. Any simple modification, change and equivalent structural change made according to the technical essence of the present application to the above embodiment are still within the protection scope of the technical solution of the present application.

Claims

1. A method for fitting irregular tunnel wall surfaces, characterized in that, Includes the following steps: Step 1: Offline acquisition of the laser pixel plane; Step 101: Use a scanner device composed of 3D LiDAR and high-precision inertial navigation to statically measure the data of the excavated tunnel wall, and save the 3D LiDAR data and pose data. Step 102: In the initial frame space coordinate system, perform point cloud stitching based on the 3D LiDAR data and pose data; Step 103: Extract the point cloud within the set range and generate an effective laser data matrix; Step 104: Based on the effective laser data matrix, fit the initial value of the planar model, and solve the optimal planar model using the optimization method to establish the laser fitting plane coordinate system; Step 105: Project the optimal spatial plane of the first frame onto the planar coordinate system below the plane to generate an effective laser pixel plane; Step 106: Take the sum of the absolute values ​​of the corresponding pixel values ​​of the ultrasonic array as the error, traverse the effective laser pixel plane, find the maximum error and save it; Step 2: Measure the normal vector and distance of the tunnel wall in real time; Step 201: Load the effective laser pixel plane; Step 202: Acquire ultrasonic array data in real time, use the RANSAC algorithm to remove outliers, and generate valid ultrasonic array data; Step 203: When the number of inner points is greater than or equal to 10, proceed to the next step; when the number of inner points is less than 10, adjust the position of the ultrasonic array and repeat steps 202 to 203. Step 204: Fit the spatial plane using the LM algorithm in the ultrasonic array coordinate system; Step 205: Calculate the pose relationship between the real-time fitting plane and the effective laser pixel plane, and project the spatial plane onto the spatial plane coordinate system; Step 206: Calculate the error between the real-time acquired value and the corresponding effective laser pixel plane value. Traverse the effective laser pixel plane with a horizontal and vertical interval set value P, repeat step 205, record the error each time, and calculate the average error. Step 207: When the average error is less than the maximum error described in step 106, calculate and save the normal vector of the spatial plane in step 204; when the average error is greater than or equal to the maximum error described in step 106, perform Gaussian sampling on the spatial plane in step 204 to generate a new fitting plane and calculate the distance from the center of the ultrasonic array to the plane.

2. The method for fitting irregular tunnel walls according to claim 1, characterized in that, The specific process of fitting the initial values ​​of the plane model in step 104 includes: Step A1: Create a planar model; A L x+B L y+C L z+D L =0 Among them, (A) L B L C L D L (x, y, z) are the parameters of the plane model to be determined, and (x, y, z) are the plane point variables; Step A2: In the point cloud within the set range, randomly select three points and substitute them into the model to calculate the model parameters; Step A3: Traverse the remaining points in the point cloud within the set range and calculate the distance from each point to the plane; Step A4: Define points whose distance from the point to the surface is less than the value M as interior points, and points whose distance from the point to the surface is greater than or equal to the value M as exterior points; Step A5: Update the planar model; Step A6: Repeat steps A1 to A4. If the number of interior points is greater than the number of interior points in the previous loop, continue to update the planar model; otherwise, save the model parameters.

3. The method for fitting irregular tunnel walls according to claim 2, characterized in that, The specific process of the optimization method described in step 104 includes: Step B1: Establish an optimization model; Where (A,B,C,D) are the parameters to be optimized, (x i ,y i ,z i Let be the 3D coordinates of the i-th point in the point cloud, and N be the number of points in the point cloud; Step B2: Using the gradient descent method, with the model parameters from step A6 as initial values, solve for the optimal planar model parameters (A). * B * C * D * ).

4. The method for fitting irregular tunnel walls according to claim 1, characterized in that, The effective laser pixel plane in step 105 has 2000*100 pixels, the horizontal and vertical spacing of the pixels corresponds to an actual distance of 1cm, and each pixel includes the vertical coordinate under the laser fitting plane coordinate system.

5. A method for fitting irregular tunnel walls according to claim 1, characterized in that, The set value P mentioned in step 206 is 5cm.

6. The method for fitting irregular tunnel walls according to claim 1, characterized in that, The specific process of Gaussian sampling described in step 207 includes: Step C1: Generate random numbers (μ1, μ2) within the range [0, 1], and generate random parameters (ΔA, ΔB): Step C2: Calculate the new plane parameters based on (ΔA, ΔB): Step C3, As a new plane parameter, it fits the spatial plane.

7. A method for fitting irregular tunnel walls according to claim 1, characterized in that, The ultrasonic array is arranged with 12 ultrasonic sensors evenly distributed on a 100cm*50cm plane, 4 horizontally and 3 vertically, with a horizontal and vertical spacing of 25cm between the sensors.

Images