A method for making mining area maps based on user-defined mapping tools

Abstract

The present application discloses a method for making a mining area map based on a user-defined mapping tool, which relates to the field of map making, including: obtaining mining area terrain and vehicle specification parameters to generate a mining area road network model; then using vehicle-mounted sensors to collect vehicle driving trajectories, slope information and point cloud data, and performing data preprocessing; then formulating a mapping standard containing mining area map element types, expressions and usage scenarios based on the collected data; finally, based on the mapping standard, extracting mining area map elements through a customized mapping tool to generate a mining area map that meets the needs of unmanned vehicles. In view of the problem of poor adaptability of mining area map making in the prior art, the present application generates a mining area map that meets the needs of unmanned vehicles.

Description

Technical Field

[0001] The present application relates to the field of map making, and in particular to a method for making a mining area map based on a user-defined mapping tool. Background Art

[0002] With the rapid development of unmanned driving technology, mining areas have gradually become one of the key scenarios for unmanned driving applications. As the core component of the unmanned driving system, high-precision maps play a key role in mining environment perception, positioning and navigation. In order to meet the needs of unmanned driving for mining area maps, various mining area high-precision map production tools continue to emerge.

[0003] Most of the existing high-precision mapping tools for mining areas are developed in a customized way, with corresponding drawing and editing functions designed for specific map elements of underground mines or open-pit mines. On the one hand, this customized design enables mapping tools to closely fit specific application scenarios, with intuitive interface layout and function buttons, and simple and efficient operation; on the other hand, with the increasing diversification of unmanned driving scenarios in mining areas, map elements are constantly changing and expanding, and customized mapping tools are difficult to flexibly adapt to element updates. Each new type of element requires the drawing function of the tool to be upgraded from the bottom layer, which has high development costs and long cycles, affecting the production efficiency and real-time performance of map data, and thus bringing safety risks to the unmanned driving system.

[0004] In addition, most of the existing mining area mapping tools use relational databases to define data models, and design tables according to the corresponding map element entities in specific business scenarios. Each element entity corresponds to a table, and each table stores its own geometry and attribute information. Although this data organization method is logically clear, it lacks abstraction and clustering of the commonality of elements, resulting in redundancy in attribute fields of similar elements. Data redundancy further exacerbates the problem of repeated development of mapping tools.

[0005] According to the standards and specifications of mining area maps, mining area map elements usually include POI points, lane center lines, stop lines, loading areas, unloading areas, parking lots, right of way relationships, etc. Developing corresponding drawing and editing functions for each type of element separately will inevitably lead to repeated development of common functions for similar elements, increasing development workload and maintenance costs. In addition, the frequent upgrade of map elements will also bring about a chain reaction of mapping tools, map SDKs, cloud, and vehicle-side adaptation updates, which will have a wide impact and a long process, restricting the production efficiency and timeliness of map data. Summary of the invention

[0006] In response to the problem of poor adaptability of mine map production in the prior art, the present application provides a mine map production method based on user-defined mapping tools. First, the mine terrain and vehicle specification parameters are obtained to generate a mine road network model; then, the vehicle driving trajectory, slope information and point cloud data are collected using on-board sensors, and data preprocessing is performed; then, based on the collected data, a mapping standard is formulated that includes mine map element types, expressions and usage scenarios; finally, based on the mapping standard, the mine map elements are extracted through a customized mapping tool to generate a mine map that meets the use needs of unmanned vehicles.

[0007] The purpose of this application is achieved through the following technical solutions.

[0008] The present application provides a method for making a mine map based on a user-defined mapping tool, comprising: obtaining the mine terrain and vehicle specification parameters, and generating a mine road network according to the mine terrain and vehicle specification parameters; using an inertial measurement unit IMU and a dynamic positioning system RTK to collect the vehicle's driving trajectory and slope information on the mine road network, and converting the collected slope information into slope information in a forward driving state of the vehicle, wherein the forward driving refers to a driving state in which the vehicle's forward direction is consistent with the lane direction; using a vehicle-mounted laser radar to collect three-dimensional point cloud data of the mine, and performing denoising and clipping preprocessing on the collected point cloud data; formulating a mine map production standard including mine map element types and corresponding expressions and usage scenarios according to the vehicle's driving trajectory data, the converted slope information and the preprocessed point cloud data; extracting mine map elements from the preprocessed point cloud data according to the set mine map production standard; the mine map elements include points of interest POI points, lane center lines, stop lines, loading areas, unloading areas, parking lots and road right relationships; wherein the road right relationship refers to the passage and locking relationship between lanes; extracting the extracted mine map elements according to Classify the geometric and attribute features according to points, lines, surfaces and relationships, and set the geometric expression and attribute information of each type of mining map elements according to the set mining map production standards; pre-set the basic map element drawing function of points, lines, surfaces and relationships in the mapping tool; based on the classified mining map elements, copy and transplant the basic map element drawing function through computer programming, and construct a mining map element drawing module that is compatible with the geometric expression and attribute information of each type of mining map element; wherein, the basic map element drawing function includes geometric drawing, geometric editing, attribute editing, rendering display and element layer management; integrate the constructed mining map element drawing module into the mapping tool; for different mining projects, if the definition of the same mining map element is different, the corresponding mining map element drawing module is labeled according to the project, and the access rights of different project users to the mining map element drawing module are controlled through user authority management; using the integrated mapping tool, according to the mining map requirements of the project, call the mining map element drawing module under the corresponding project label, draw the mining map elements, and generate the mining map.

[0009] Furthermore, the terrain and vehicle specification parameters of the mining area are obtained, and a mining area road network is generated according to the terrain and vehicle specification parameters of the mining area, including: obtaining terrain data of the mining area, the terrain data of the mining area includes the geometric parameters of the mining area roads and the distribution information of the mining area objects; obtaining vehicle specification parameters of the mining area, the vehicle specification parameters include the size, load and power performance of the vehicle; constructing a three-dimensional terrain model of the mining area based on the obtained terrain data of the mining area, and identifying the primitives including the roads and objects in the three-dimensional terrain model of the mining area; based on the obtained vehicle specification parameters of the mining area, simulating the motion characteristics of the vehicle by vehicle dynamics simulation in the three-dimensional terrain model of the mining area, and obtaining the driving ability parameters of the vehicle on the mining area roads; constraining the width, slope and turning radius of the mining area roads according to the driving ability parameters, and generating a mining area road network model that meets the vehicle traffic requirements.

[0010] Furthermore, the inertial measurement unit IMU and the dynamic positioning system RTK are used to collect the driving trajectory and slope information of the vehicle on the mining area road network, and the collected slope information is converted into the slope information when the vehicle is in the forward driving state, including: using the inertial measurement unit IMU to collect the posture information of the vehicle on the mining area road network, and using the dynamic positioning system RTK to collect the position information of the vehicle on the mining area road network, and calculating the driving trajectory of the vehicle on the mining area road network according to the collected posture information and position information; wherein the posture information includes the vehicle's yaw angle, pitch angle and roll angle, and the position information includes the vehicle's latitude and longitude coordinates and elevation; using the inertial measurement unit IMU to collect the acceleration information of the vehicle during driving , according to the collected acceleration information, the acceleration vector decomposition method is used to calculate the slope information of the vehicle relative to the horizontal plane; the vehicle's heading angle information and the lane direction information are obtained, and the angle between the vehicle's heading angle and the lane direction is calculated. When the absolute value of the angle is less than or equal to the preset threshold, the vehicle is judged to be in a forward driving state, and the calculated slope information remains unchanged; when the absolute value of the angle is greater than the preset threshold, the vehicle is judged to be in a non-forward driving state, and the calculated slope information is multiplied by -1 to obtain the corresponding slope information under the forward driving state of the vehicle; wherein, the vehicle's heading angle information is obtained through the vehicle-mounted inertial navigation system or the vehicle-mounted electronic compass, and the lane direction information is extracted from the calculated vehicle driving trajectory.

[0011] Furthermore, an inertial measurement unit IMU is used to collect the posture information of the vehicle on the mining area road network, and a dynamic positioning system RTK is used to collect the position information of the vehicle on the mining area road network. According to the collected posture information and position information, the driving trajectory of the vehicle on the mining area road network is calculated, including: using the inertial measurement unit IMU to continuously collect the angular velocity and acceleration information of the vehicle at a preset sampling frequency, and using the dynamic positioning system RTK to continuously collect the position information of the vehicle at a preset sampling frequency, and synchronously obtaining the timing information of the vehicle posture and position; wherein the sampling frequency of the IMU is higher than the sampling frequency of the RTK, and the information collected by the two sensors is aligned by time stamp; using a zero-speed correction algorithm and a static alignment algorithm, the collected angular velocity and acceleration information are corrected and aligned to eliminate the zero bias error and installation error of the IMU The error is obtained to obtain the corrected angular velocity and acceleration information; the Kalman filter algorithm is used to fuse the corrected angular velocity and acceleration information with the position information collected by RTK to estimate the vehicle's posture and position state, and obtain the vehicle's posture information and position information; according to the vehicle's posture information, the Euler angle conversion formula is used to convert the speed information in the vehicle coordinate system to the mining area road network coordinate system, and the speed information of the vehicle in the mining area road network coordinate system is obtained; according to the vehicle's position information, the coordinate conversion formula is used to convert the vehicle's latitude and longitude coordinates and elevation to the mining area road network coordinate system, and the position information of the vehicle in the mining area road network coordinate system is obtained; the Runge-Kutta integration algorithm is used to integrate the obtained speed information over time, and superimpose it with the obtained position information to obtain the vehicle's driving trajectory in the mining area road network coordinate system.

[0012] Furthermore, the obtained speed information is integrated with respect to time using the Runge-Kutta integration algorithm, and is superimposed with the obtained position information to obtain the vehicle's driving trajectory in the mining area road network coordinate system, including: recording the vehicle's speed information in the mining area road network coordinate system as , the location information is recorded as ; According to the Runge-Kutta integration algorithm, the time t is discretized into , the time step is h, then ; and speed and location In time The values ​​at are recorded as and ; Use the fourth-order Runge-Kutta integral formula to calculate the speed The integral increment within the time step h : ; ; ; ; ;in, Indicates at time ,Location The speed value at Indicates at time ,Location The speed value at the point; and so on; by interpolating the speed information from adjacent moments Calculated; using location With speed integral increment , iteratively calculate the position at the next moment : ; Repeat the iteration until the final moment is calculated Location , obtained by to The vehicle driving trajectory formed; in, For the initial moment location, Indicates the time interval Internal speed Perform a definite integral.

[0013] Furthermore, a vehicle-mounted laser radar is used to collect three-dimensional point cloud data of the mining area, and the collected point cloud data is denoised and cropped for preprocessing, including: using the vehicle-mounted laser radar to collect three-dimensional point cloud data of the mining area at a preset scanning frequency and scanning angle to obtain original point cloud data reflecting the topographical features of the mining area; denoising the original point cloud data: calculating the density of each point in the original point cloud data, marking the points with a density less than a threshold as noise points, and removing the noise points using a statistical filtering method to obtain point cloud data with uniform density; using an octree algorithm to voxelize the point cloud data with uniform density, merging the points in each voxel, and replacing the original points with the center points of the voxels to obtain point cloud data with reduced density; using a Laplace operator to calculate the curvature of each point in the point cloud data with reduced density, performing Laplace smoothing on the points with a curvature greater than a threshold to obtain point cloud data with a smooth surface as denoised point cloud data; and cropping the denoised point cloud data to obtain preprocessed point cloud data.

[0014] Furthermore, the denoised point cloud data is cropped, including: spatially aligning the denoised point cloud data with the obtained vehicle driving trajectory, setting a preset spatial range with the vehicle driving trajectory as the center, using the spatial range to spatially crop the point cloud data, and extracting local point cloud data related to the vehicle driving path; setting a preset angle range based on the slope information obtained in the forward driving state of the vehicle, and using the angle range to perform angle cropping on the extracted local point cloud data to obtain local point cloud data related to the vehicle driving slope as preprocessed point cloud data for mining area map production.

[0015] Furthermore, based on the vehicle driving trajectory data, the converted slope information and the pre-processed point cloud data, a mining area map production standard including mining area map element types and corresponding expression methods and usage scenarios is formulated, including: determining the mining area map element types to be expressed including roads, slopes and terrain features based on the obtained vehicle driving trajectory data, the obtained vehicle slope information and the obtained pre-processed point cloud data, and determining the usage scenarios of each element type based on the use requirements of unmanned vehicles in the mining area; formulating corresponding element expression standards based on each element type: for road elements, the vehicle driving trajectory is used to extract the road centerline, and the road centerline is used as the skeleton, and the road centerline is used as the skeleton. The road boundary is extracted using preprocessed point cloud data, expressed as a set of directed line segments, and the road width and length attributes are annotated. For slope elements, the longitudinal and lateral slope values ​​of the road are calculated using vehicle slope information and preprocessed point cloud data. The slope values ​​are divided into different levels, and different levels of slope are represented by different colors or symbols, which are superimposed on the road elements. For terrain and object elements, the terrain undulation information and object distribution information within the preset range of the road are extracted using preprocessed point cloud data, and expressed as different geometric primitives according to the characteristics of the terrain and objects. The set element expression standard is associated with the determined element type and the corresponding usage scenario as the standard for mining area map production.

[0016] Furthermore, according to the set mining area map production standards, mining area map elements are extracted from the preprocessed point cloud data, including: according to the established mining area map production standards, mining area map elements are extracted from the obtained preprocessed point cloud data: the preprocessed point cloud data are classified using a point cloud semantic segmentation method based on deep learning, and point cloud subsets of different categories of roads, stop lines, loading areas, unloading areas and parking lots are extracted; Euler clustering is performed on the extracted road point cloud subsets to obtain road centerline point clouds, and the road centerline point clouds are fitted using the least squares method to extract lane centerline elements; plane fitting is performed on the point cloud subsets of the loading area, unloading area and parking lot functional areas, and plane boundary lines are extracted to obtain vectorized representations of the elements of each functional area; feature points are extracted from the point cloud data using a key point detection-based method, and the feature points are used as point of interest (POI) elements; a topological connectivity graph is constructed based on the extracted lane lines, and the road right relationship between lanes is determined by analyzing the topological connectivity graph.

[0017] Further, a mining area map element drawing module is constructed, including: for point-type mining area map elements, the geometric drawing and attribute editing functions of the point elements are copied, and a point of interest POI drawing module is constructed, and the location information of the point of interest POI is input into the drawing function to realize the geometric drawing of the point of interest POI, and the name, type and other attribute information of the point of interest POI are associated with the drawn point element; for line-type mining area map elements, the geometric drawing and attribute editing functions of the line elements are copied, and a lane centerline and stop line drawing module is constructed, and the location and direction information of the lane centerline and stop line are input into the drawing function to realize their geometric drawing, and the attribute information such as the width, length and driving direction of the lane centerline and the stop line, as well as the type information of the stop line are associated with the drawn line element; for area-type For mining map elements of the right-of-way relationship type, copy the geometric drawing and attribute editing functions of the surface elements, build drawing modules for loading areas, unloading areas and parking lots, input the boundary line information of each functional area into the drawing function to realize its geometric drawing, and associate the material type of the loading area and unloading area and the number of parking spaces, entrance and exit locations and other attribute information with the drawn surface elements; for mining map elements of the right-of-way relationship type, copy the attribute editing function of the relationship elements, build a right-of-way relationship drawing module, geometrically express the right-of-way relationship through directed line segments, and associate attribute information such as travel direction, turn type and priority with the drawn relationship line segments; transplant the rendering display and feature layer management functions to each mining map element drawing module to realize the visual display and layer management of mining map elements.

[0018] Furthermore, the integrated mapping tool is used to create or select a project label corresponding to the mining area map to be produced; under the selected project label, the corresponding mining area map element drawing module is called from the mapping tool interface according to the requirements of the mining area map; specifically, if it is necessary to draw a point of interest POI, the point of interest POI drawing module is called; if it is necessary to draw a lane centerline or a stop line, the lane centerline drawing module or the stop line drawing module is called; if it is necessary to draw a loading area, an unloading area or a parking lot, the loading area drawing module, the unloading area drawing module or the parking lot drawing module is called; if it is necessary to edit the road right relationship, the road right relationship drawing module is called. In the called mining area map element drawing module, the geometric shapes of mining area map elements are drawn on the map by means of human-computer interaction using a mouse or other input devices; after the drawing is completed, the attribute information of the drawn mining area map elements is entered into the drawing module, including name, type, description, etc.; the drawn mining area map elements are rendered and displayed, and added to the map in the form of layers, and the display and hiding of the feature layers can be controlled by the feature layer management function in the drawing module; the execution is repeated until all required mining area map elements are drawn; and a complete mining area map is automatically generated based on the drawn mining area map elements.

[0019] Compared with the prior art, the advantages of this application are:

[0020] By obtaining data such as mining area terrain, vehicle specifications, vehicle driving trajectory, slope, and point cloud, and formulating mining area map production standards based on the use requirements of unmanned vehicles, the generated mining area map can better reflect the actual conditions of mining area roads and environment, and meet the use requirements of unmanned vehicles for high-precision maps. Introducing the actual terrain and vehicle parameters of the mining area into the map production process can improve the adaptability of the map to the mining area environment. Combining vehicle driving data with mapping requirements ensures the fit between the map content and the unmanned driving application scenario, thereby significantly improving the practicality and effectiveness of the mining area map.

[0021] By abstracting mining map elements such as roads, slopes, and landforms into geometric and attribute features, and designing unified expression standards and drawing methods, the presentation of different map elements is standardized. Unified element expression is conducive to the accurate identification and extraction of map information by the unmanned driving system. The vectorized representation of road centerlines, functional area boundaries, and the definition of key point POIs improve the geometric description accuracy of map elements and enhance the semantic characteristics of the map. The use of intuitive visual elements such as color and symbols to distinguish different slope levels optimizes the visual presentation of map elements, making complex environmental information clearer and easier to read. The standardization and refinement of map element representation ensures the consistency of mining area maps in different projects and scenarios, and improves map quality.

[0022] The expression of mining map elements such as roads, slopes, and landforms is automatically converted into drawing modules through computer programming and integrated into user-defined mapping tools. Compared with manual drawing, automated feature extraction and symbolic drawing greatly shorten the mapping time. The conversion of mapping knowledge to mapping tools is realized through programming, so that cartographers do not need to manually modify drawing instructions, but only need to call the corresponding functional modules to quickly complete the extraction and production of map elements. At the same time, map elements are extracted based on sensor data, which reduces the dependence on manual annotation. The data-driven mapping method can flexibly adapt to the map update and re-production needs of different mining areas.

[0023] The project management mechanism is introduced to customize the map element drawing module for different mining areas. Through the project tag setting, the flexible combination of layer configuration and symbolization parameters is realized to meet the personalized mapping needs of different projects. While sharing mapping resources, the user authority control is also used to avoid the misoperation of irrelevant personnel. The new mapping requirements are converted into code-level updates without modifying the mapping process from the bottom up, which greatly improves the scalability and maintainability of mapping tools.

[0024] The vehicle trajectory and posture data collected by GPS / IMU / RTK positioning equipment and the three-dimensional point cloud data scanned by LiDAR are comprehensively used to reflect the mining environment in all aspects; through trajectory and point cloud registration, slope transformation, point cloud filtering and other data preprocessing and fusion methods, the spatial consistency and signal-to-noise ratio of vehicle-mounted sensor data are optimized, and refined mining map elements are extracted in combination with vehicle and road characteristics. The fusion of multi-source data expands the dimension of map information, makes up for the shortcomings of a single sensor in obtaining global environmental information, and quantifies the relationship between the vehicle and the environment into measurable map parameters, so that the final constructed mining map not only describes the macro characteristics of the mining environment, but also presents the local road conditions from the perspective of vehicle driving, greatly improving the precision of the map. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The present application will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0026] Figure 1 is an exemplary flow chart of a method for making a mine map based on a user-defined mapping tool according to some embodiments of the present application;

[0027] Figure 2 is an exemplary flow chart for obtaining vehicle driving trajectory and slope information according to some embodiments of the present application;

[0028] Figure 3 is an exemplary flow chart of constructing a mining area road network model according to some embodiments of the present application; DETAILED DESCRIPTION

[0029] The method and system provided in the embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0030] Figure 1This is an exemplary flow chart of a method for making a mine map based on a user-defined mapping tool according to some embodiments of the present application, including: obtaining the mine terrain and vehicle specification parameters, and generating a mine road network according to the mine terrain and vehicle specification parameters; using an inertial measurement unit IMU and a dynamic positioning system RTK to collect the vehicle's driving trajectory and slope information on the mine road network, and converting the collected slope information into the slope information of the vehicle in the forward driving state, where the forward driving direction of the vehicle is consistent with the lane direction; using a vehicle-mounted laser radar to collect a three-dimensional image of the mine area Point cloud data, and denoise and crop the collected point cloud data for preprocessing; formulate a mining map production standard that includes mining map element types and corresponding expressions and usage scenarios based on vehicle driving trajectory data, converted slope information and preprocessed point cloud data; extract mining map elements from the preprocessed point cloud data according to the set mining map production standard; mining map elements include points of interest (POI), lane center lines, stop lines, loading areas, unloading areas, parking lots and road rights; among them, road rights refer to the passage and locking relationship between lanes; extract The mining map elements are classified according to the geometric and attribute characteristics of points, lines, surfaces and relationships, and the geometric expression and attribute information of each type of mining map elements are set according to the set mining map production standards; the basic map element drawing functions of points, lines, surfaces and relationships are pre-set in the mapping tool; based on the classified mining map elements, the basic map element drawing functions are copied and transplanted through computer programming to construct a mining map element drawing module that is compatible with the geometric expression and attribute information of each type of mining map element; wherein the basic map element drawing function includes geometric drawing, geometric editing, attribute editing, rendering display and element layer management; the constructed mining map element drawing module is integrated into the mapping tool; for different mining projects, if the definition of the same mining map element is different, the corresponding mining map element drawing module is labeled according to the project, and the access rights of different project users to the mining map element drawing module are controlled through user authority management; using the integrated mapping tool, according to the mining map requirements of the project, the mining map element drawing module under the corresponding project label is called to draw the mining map elements and generate the mining map.

[0031] Data input, obtain mining area terrain data: obtain the geometric parameters of mining area roads from the mining area geographic information system or database, such as road width, length, curvature, slope, etc.; obtain the distribution information of land objects in the mining area, such as the location and size of buildings, obstacles, etc.; obtain mining area vehicle specification parameters: obtain the vehicle's size parameters from the vehicle management system or database, such as length, width, height, wheelbase, etc.; obtain the vehicle's load parameters, such as empty weight, full load weight, etc.; obtain the vehicle's power performance parameters, such as engine power, torque, gearbox ratio, etc.

[0032] Collect data through on-board sensors: Install an inertial measurement unit (IMU) sensor to collect vehicle posture information. The IMU continuously collects the vehicle's angular velocity and acceleration information at a preset sampling frequency (such as 100Hz); the angular velocity information includes the vehicle's rotation speed around the three axes of roll, pitch, and yaw; the acceleration information includes the vehicle's acceleration in the three directions of front and back, left and right, and up and down. Install a dynamic positioning system (RTK) receiver to collect the vehicle's position information. RTK continuously collects the vehicle's latitude and longitude coordinates and elevation information at a preset sampling frequency (such as 10Hz).

[0033] Synchronize the data collected by IMU and RTK through timestamp: record the start time of data collection and sampling frequency , ; For the i-th data point of the IMU, its timestamp is ; For the jth data point of RTK, its timestamp is ; For each ,exist and Linear interpolation is performed between the two values ​​to obtain the position information at the corresponding time. ; IMU data and the corresponding location information Combined into timing information .

[0034] Use the zero-speed correction algorithm to correct the IMU data: Determine the moment when the vehicle is stationary: When the acceleration and angular velocity amplitudes of the IMU are lower than the preset threshold, the vehicle is considered stationary; collect the IMU data when the vehicle is stationary, and calculate the average of the acceleration and angular velocity and ; IMU data during motion , minus the bias error: .

[0035] The IMU data is aligned using a static alignment algorithm. When the vehicle is stationary, the installation deflection angle of the IMU relative to the vehicle coordinate system is calculated by least squares estimation. : Collect IMU data when the vehicle is stationary, including acceleration data and angular velocity data Based on the known reference information, construct the observation equation: The theoretical value of gravity acceleration in the vehicle coordinate system (Assume that the vehicle is stationary on a horizontal plane); the observation equation is: ;in, is the rotation matrix from the IMU coordinate system to the vehicle coordinate system, is the observation noise of the accelerometer.

[0036] Substitute the collected data at multiple stationary moments into the observation equation to construct the least squares problem: ,in, is the rotation matrix to be optimized; is the kth static acceleration measurement; is the gravity vector.

[0037] Use an optimization algorithm (such as the Levenberg-Marquardt algorithm) to solve the above least squares problem and get the estimated rotation matrix ; Based on the estimated rotation matrix , calculate the installation deflection angle of the IMU relative to the vehicle coordinate system ; Euler angles can be used Or quaternion To represent. For the IMU data imu_i_corrected during the motion process, alignment compensation is performed: for the IMU data imu_i_corrected at each moment (which has been corrected for zero bias), the estimated installation deflection error_align is used for alignment compensation. The compensation process can be achieved through rotation transformation: ; Where f represents the function of rotating the IMU data according to the installation deflection angle; Represents the aligned IMU data; Represents the corrected IMU data; Indicates alignment error.

[0038] Specifically, Convert to rotation matrix form ; Rotate the acceleration data: ; Perform a rotation transformation on the angular velocity data: ; Combine the aligned acceleration and angular velocity data into .

[0039] Using the Kalman filter algorithm, the corrected angular velocity and acceleration information are fused with the position information collected by RTK to estimate the vehicle's attitude and position state, and establish the state equation and observation equation of the Kalman filter. The state vector includes the vehicle's position, velocity, attitude and other information: The composition of the state vector X is: ; p is the position of the vehicle in the navigation coordinate system (such as the ENU coordinate system), including longitude, latitude, and elevation; v is the speed of the vehicle in the navigation coordinate system, including eastward speed, northward speed, and radial speed; q is the vehicle's attitude quaternion, which represents the rotation relationship between the vehicle coordinate system and the navigation coordinate system; state equation: ; f is the state transfer function, which describes the change law of the state vector X within the time step; u is the control input, which can be the angular velocity and acceleration information provided by the IMU; w is the process noise, which represents the uncertainty in the state prediction process, and the observation equation is: , h is the observation function, which maps the state vector X to the observation space. The observation here can be the position information provided by RTK, and v is the observation noise, which represents the uncertainty in the observation process. According to the angular velocity and acceleration information provided by IMU, the attitude and position state of the vehicle are predicted through attitude solution and velocity integration: Attitude solution: Using angular velocity information, attitude prediction is performed through quaternion differential equations. , where q is the attitude quaternion and ω is the angular velocity vector (converted to quaternion form).

[0040] Use numerical integration methods (such as Euler method, Runge-Kutta method) to solve the quaternion differential equation and obtain the predicted attitude quaternion , Velocity integral: Use acceleration information to predict velocity using Newton’s laws of motion, , where v is the velocity vector and a is the acceleration vector (the influence of gravity acceleration needs to be considered). The velocity differential equation is solved using the numerical integration method to obtain the predicted velocity vector Position integral: Using the predicted velocity information, the position increment is calculated by definite integral. , where p is the position vector; the position differential equation is solved using the numerical integration method to obtain the predicted position vector ; Update the observation equation of the Kalman filter based on the position information provided by RTK: use the position information (longitude, latitude, elevation) provided by RTK as the observation vector ; According to the observation function h, the predicted state vector Mapped to the observation space, we get the predicted observation vector . Compute the observation residuals: , based on the observed residuals and the Kalman gain matrix, update the estimated value of the state vector: ; Where K is the Kalman gain matrix, which is calculated based on the state covariance matrix and the observation noise covariance matrix; The Kalman filter algorithm is used to estimate the optimal attitude and position state of the vehicle, and the filtered attitude information (Euler angle or quaternion) and position information (latitude, longitude, and elevation) are obtained: The prediction and update steps are iteratively performed to continuously update the estimated value of the state vector ; From the state vector The vehicle's position information (longitude, latitude, elevation) and attitude information (Euler angle or quaternion) are extracted from the Kalman filter algorithm to obtain the optimal estimated value after filtering.

[0041] Figure 2 This is an exemplary flow chart for obtaining vehicle driving trajectory and slope information according to some embodiments of the present application, converting the speed information in the vehicle coordinate system to the mine road network coordinate system, and using the fourth-order Runge-Kutta integral algorithm to calculate the vehicle driving trajectory, and according to the vehicle posture information, using the Euler angle conversion formula, converting the speed information in the vehicle coordinate system to the mine road network coordinate system: assuming that the vehicle coordinate system is the front right lower (FRD) coordinate system, and the mine road network coordinate system is the northeast celestial (ENU) coordinate system; constructing a rotation matrix from the vehicle coordinate system to the mine road network coordinate system based on the posture information (Euler angle or quaternion) estimated by the Kalman filter If the attitude information is Euler angle (heading angle , pitch angle θ, roll angle ), then the rotation matrix It can be expressed as: ;in, are the rotation matrices for rotation around the x-axis, y-axis, and z-axis respectively.

[0042] If the attitude information is a quaternion , then the rotation matrix It can be expressed as:

[0043] Among them, q1 to q4 represent the components of the quaternion.

[0044] The velocity vector in the vehicle coordinate system To perform coordinate transformation: ,in, ; Get the vehicle speed information in the mining area road network coordinate system .

[0045] According to the vehicle's location information, the vehicle's latitude and longitude coordinates and elevation are converted to the mine road network coordinate system using the coordinate conversion formula: According to the origin of the mine road network coordinate system (longitude ,latitude , elevation ) and the scaling ratio scale, establish the conversion relationship between longitude and latitude coordinates and mining area road network coordinates; longitude difference , latitude difference , elevation difference ; Easting coordinate , north coordinate , celestial coordinates ; The latitude and longitude coordinates of the vehicle Perform coordinate conversion to obtain the vehicle's position information in the mining area road network coordinate system .

[0046] The fourth-order Runge-Kutta integration algorithm is used to integrate the speed information with respect to time and superimpose it with the position information to obtain the vehicle's driving trajectory: time t is discretized into , the time step is h, then ; Set the speed and location The values ​​at time ti are recorded as and ; Use the fourth-order Runge-Kutta integral formula to calculate the speed The integral increment within the time step h : ; ; ; ; ;in, Indicates at time ,Location The speed value at the adjacent time is obtained by interpolation. Calculated. Using location With speed integral increment , iteratively calculate the position at the next moment : ; Repeat the iteration until the final moment is calculated Location , obtained by to The vehicle driving trajectory.

[0047] The inertial measurement unit (IMU) is used to collect acceleration information, and the slope information of the vehicle in the forward driving state is calculated in combination with the vehicle heading angle and lane direction information. The inertial measurement unit (IMU) is used to collect the acceleration information of the vehicle during driving: IMU usually contains a three-axis accelerometer, which can measure the acceleration of the vehicle in the front and back, left and right, and up and down directions; the IMU is fixed on the vehicle so that its coordinate axis is aligned with the vehicle coordinate system; during vehicle driving, the IMU continuously collects acceleration information at a certain sampling frequency (such as 100Hz); the collected acceleration information is recorded to obtain a series of acceleration sample data.

[0048] The acceleration vector decomposition method is used to calculate the slope information of the vehicle relative to the horizontal plane, and the collected acceleration sample data is preprocessed: Remove zero bias: Since the accelerometer has a zero-point offset, that is, the acceleration value output in a stationary state is not zero, the collected acceleration data needs to be compensated for zero bias; collect acceleration data for a period of time when the vehicle is stationary, and calculate the average value of each axis as the zero bias estimate; subtract the zero bias estimate of the corresponding axis from the collected acceleration data to obtain the acceleration data after removing the zero bias. Correct the scale factor: Since there is an error in the sensitivity of the accelerometer, the collected acceleration data needs to be corrected for the scale factor; use the known gravity acceleration value (such as ) calibrate the accelerometer to obtain the scale factor of each axis; divide the acceleration data after removing the zero bias by the scale factor of the corresponding axis to obtain the corrected acceleration data.

[0049] Perform sliding average or low-pass filtering on the processed acceleration data: Sliding average: perform arithmetic averaging on N consecutive acceleration samples to obtain the smoothed acceleration value; set the size of the sliding window N, determined according to the sampling frequency and the required smoothing degree; perform sliding average on the acceleration data of each axis to obtain the smoothed acceleration data; Low-pass filtering: use a digital low-pass filter to filter the acceleration data to remove high-frequency noise; design a suitable low-pass filter, such as Butterworth filter, Kalman filter, etc.; perform low-pass filtering on the acceleration data of each axis to obtain the filtered acceleration data. Using the acceleration vector decomposition method, decompose the acceleration vector in the vehicle coordinate system into the horizontal and vertical directions: the acceleration vector in the vehicle coordinate system is recorded as ; Assume that the x-axis of the vehicle coordinate system is consistent with the vehicle's forward direction, and the z-axis is vertically upward. The component of acceleration on the horizontal plane It can be calculated by taking the square root of the sum of the squares of the x-axis and y-axis accelerations: The vertical component of acceleration That is the z-axis acceleration: ; Calculate the slope angle θ of the vehicle relative to the horizontal plane: Use the inverse tangent function Calculate the slope angle θ: ;in, The function can be based on and The sign of determines the quadrant of θ. The θ value returned by the function is in radians, which can be converted to degrees: If you need to express the slope angle as a percentage, you can calculate the tangent and multiply it by 100%: . Get the slope information of the vehicle relative to the horizontal plane: slope angle Indicates the vehicle's inclination angle relative to the horizontal plane, in degrees; slope percentage Indicates the slope of the vehicle relative to the horizontal plane, in percentage; a positive value indicates that the vehicle is traveling uphill, and a negative value indicates that the vehicle is traveling downhill.

[0050] Extract lane direction information from the vehicle's driving trajectory and obtain the vehicle's heading angle information: Use the attitude information provided by the vehicle-mounted inertial navigation system (INS) to directly obtain the vehicle's heading angle: INS calculates the vehicle's attitude information by integrating the data of the inertial measurement unit (IMU) and the global navigation satellite system (GNSS); the attitude information output by INS includes the vehicle's heading angle, pitch angle, and roll angle; extract the heading angle from the attitude information output by INS to obtain the vehicle's current heading angle information. Use the vehicle-mounted electronic compass to measure the vehicle's angle relative to the geomagnetic North Pole to obtain the vehicle's heading angle: The vehicle-mounted electronic compass determines the vehicle's direction relative to the geomagnetic North Pole by measuring the strength and direction of the geomagnetic field at the vehicle's location; the direction information output by the electronic compass is usually given in the form of an angle, indicating the angle between the vehicle's heading and the direction of the geomagnetic North Pole; the angle output by the electronic compass is used as the vehicle's heading angle to obtain the vehicle's current heading angle information.

[0051] Extract lane direction information from the calculated vehicle trajectory: Perform local fitting on the vehicle trajectory to obtain the tangent direction at each point on the trajectory: The vehicle trajectory is usually composed of a series of discrete position points, each of which includes information such as longitude, latitude, and elevation; Perform local fitting on the position points on the trajectory to obtain a local approximate representation of the trajectory, such as a straight line segment, circular arc, etc.; For each position point on the trajectory, calculate the tangent direction on the local approximate curve where it is located; If the local approximation is a straight line segment, the tangent direction is the direction of the straight line segment; If the local approximation is a circular arc, the tangent direction is the tangent direction at the point on the circular arc, and the tangent direction at each position point on the trajectory is obtained to form a tangent direction sequence. Take the tangent direction as the lane direction and extract the lane direction information at the point closest to the current position of the vehicle: According to the current position information of the vehicle, find the position point closest to the vehicle on the trajectory, extract the corresponding tangent direction at the position point, and use it as the lane direction of the current position of the vehicle. The lane direction can be expressed in angle form, such as the angle with the true north direction, or in unit vector form.

[0052] Comprehensive use of vehicle heading angle information and lane direction information: unify the vehicle heading angle information and lane direction information into the same coordinate system, such as the geographic coordinate system or the vehicle coordinate system, and compare the angle between the vehicle heading angle and the lane direction to determine whether the vehicle is consistent with the lane direction. If the angle between the vehicle heading angle and the lane direction is small, it means that the vehicle is basically traveling along the lane direction. If the angle between the vehicle heading angle and the lane direction is large, it means that the vehicle may have deviated from the lane and needs to be corrected.

[0053] Calculate the angle between the vehicle heading angle and the lane direction to determine whether the vehicle is in a forward driving state: the vehicle heading angle is , the lane direction angle is ; Calculate the angle between the vehicle heading angle and the lane direction ; Set a preset threshold (such as 30°) is used to determine whether the vehicle is in a forward driving state. When the vehicle is judged to be in a forward driving state, the calculated slope information remains unchanged. When the vehicle is judged to be in a non-forward driving state, the calculated slope information is multiplied by -1 to obtain the corresponding slope information when the vehicle is in the forward driving state.

[0054] Figure 3 This is an exemplary flow chart for constructing a mining area road network model according to some embodiments of the present application, using a vehicle-mounted laser radar to collect three-dimensional point cloud data of the mining area: installing the vehicle-mounted laser radar on mining area operating vehicles, such as dump trucks, loaders, etc.; setting the scanning frequency and scanning angle of the laser radar to meet the needs of collecting topographic and geomorphic features of the mining area; the scanning frequency is usually set to 10-100Hz, and the scanning angle is usually set to 360° or 180°; during vehicle driving, the laser radar continuously collects three-dimensional point cloud data of the mining area at a preset scanning frequency and scanning angle; the laser radar calculates the three-dimensional coordinates (x, y, z) of each sampling point by measuring the emission and return time of the laser pulse; and obtaining original point cloud data reflecting the topographic and geomorphic features of the mining area, each point including three-dimensional coordinate information.

[0055] De-noise the original point cloud data and calculate the density of each point in the original point cloud data: For each point in the point cloud data , defining a spherical neighborhood around , the neighborhood radius is r; the statistics fall within the neighborhood The number of other points within , get the point Density value , set a density threshold , the density value is less than The points are marked as noise points, and the density threshold It can be selected according to the characteristics of the point cloud data and the noise level, and is usually taken as a certain proportion of the average density (such as 0.5 times).

[0056] Use statistical filtering method to remove noise points: perform statistical analysis on the original point cloud data and calculate the The distance and height difference to its surrounding points: For each point , calculate its relationship with the neighborhood The Euclidean distance to other points in and height difference , get the point The distance set to its surrounding points and height difference set .

[0057] According to the distance collection and height difference set The distribution of noise points is determined by calculating the distance set. The mean and standard deviation , determine the distance threshold , calculate the height difference set The mean and standard deviation , determine the height difference threshold , the parameter k can be selected according to the distribution characteristics of the noise points, usually taking a value of 2 or 3. , to determine whether it meets the conditions for removing noise points: if the point The distance between any point in its neighborhood is greater than , or the height difference is greater than , then the point Mark as noise points, remove all points marked as noise points from the point cloud data, and obtain point cloud data with uniform density.

[0058] Use the octree algorithm to voxelize the point cloud data with uniform density: construct a cubic bounding box surrounding the point cloud data and determine the minimum vertex coordinates of the bounding box and the maximum vertex coordinates . Based on the preset voxel size , calculate the number of voxels in each direction of the bounding box: number of voxels , the number of voxels , the number of voxels , divide the point cloud data into A cubic grid, each grid is called a voxel; for each voxel , calculate the center point coordinates of all points inside it , using the center point Replace voxel The original points in the voxel are processed by voxelization, and multiple points in each voxel are merged to reduce the density of point cloud data.

[0059] The curvature of each point in the point cloud data with reduced density is calculated using the Laplacian operator: For each point in the point cloud data , defining a spherical neighborhood around , the neighborhood radius is r. Calculate the point and its neighbors The Laplacian value of the local surface formed by the points in :Point Fit a local surface to the points in its neighborhood , such as fitting a quadratic surface using the least squares method, calculating the local surface On point The Laplace operator value at , reflects the curvature change of the surface at that point, and sets a curvature threshold , mark the points whose Laplace operator values ​​are greater than lap_max as points that need to be smoothed. The curvature threshold lap_max can be selected according to the surface characteristics and noise level of the point cloud data, and is usually taken as a certain multiple of the average Laplace operator value (such as 2 times).

[0060] Perform Laplace smoothing on the points that need to be smoothed: For each point that needs to be smoothed , use the Laplace smoothing algorithm to iteratively smooth it: calculate the point New coordinates of , where λ is the smoothing coefficient, which controls the intensity of smoothing and is usually between 0.1 and 0.5. Yes The normal vector at can be obtained by fitting the local surface The Laplace smoothing algorithm is iteratively executed until the stop condition is met (such as the number of iterations reaches a preset value or the point coordinate change is less than a threshold). After Laplace smoothing, the noise points in the point cloud data are smoothed, the surface becomes smoother, and the denoised point cloud data is obtained.

[0061] The denoised point cloud data is cropped and spatially aligned with the vehicle's driving trajectory: the GPS receiver on the vehicle is used to obtain the vehicle's location information, including longitude, latitude and elevation; the IMU (inertial measurement unit) on the vehicle is used to obtain the vehicle's attitude information, including roll angle, pitch angle and yaw angle. The GPS position information and IMU attitude information are synchronized in time to obtain the vehicle's position and attitude at each moment. Through time interpolation and other methods, a continuous vehicle driving trajectory is generated, which is represented as a series of position points and corresponding attitudes.

[0062] Perform coordinate transformation on the denoised point cloud data, converting it from the lidar coordinate system to the vehicle coordinate system: determine the transformation relationship between the lidar coordinate system and the vehicle coordinate system based on the installation position and angle of the lidar relative to the vehicle, apply the transformation matrix to each point in the point cloud data, and convert its coordinates from the lidar coordinate system to the vehicle coordinate system.

[0063] Through the spatial registration algorithm (such as ICP algorithm), the converted point cloud data is aligned with the vehicle's driving trajectory to the same coordinate system: select the feature points on the vehicle's driving trajectory (such as turning points, parking points, etc.) as reference points, find the feature points corresponding to the reference points in the point cloud data, establish point correspondence, and calculate the transformation matrix between the point cloud data and the vehicle's driving trajectory by minimizing the error of the point correspondence. Apply the transformation matrix to the point cloud data to align it with the vehicle's driving trajectory to the same coordinate system. Set a preset spatial range with the vehicle's driving trajectory as the center, and perform spatial cropping on the point cloud data: Set a preset spatial range with each position point on the vehicle's driving trajectory as the center, such as a spherical area with a radius of R. For each point in the point cloud data , to determine whether it falls within the spatial range of a certain position point on the vehicle's driving trajectory: Calculate point The distance to each position point on the vehicle's trajectory , if there exists a position point such that If the radius R is less than or equal to the preset spatial range, the point Mark the points to be retained.

[0064] Extract all the points marked as needing to be retained to obtain local point cloud data related to the vehicle's driving path. The selection of the spatial range radius R needs to consider factors such as the vehicle's driving speed, the scanning range of the laser radar, and the required map accuracy. It is usually 10-50 meters. Based on the slope information of the vehicle in the forward driving state, set a preset angle range, and perform angle cropping on the extracted local point cloud data: Use sensors such as IMU to obtain the vehicle's pitch angle information, indicating the slope of the vehicle in the forward driving state, and calculate the vehicle's current driving slope slope_vehicle based on the vehicle's pitch angle.

[0065] Set a preset angle range, such as ±10°, and use the vehicle's driving slope as the center to get the upper and lower boundaries of the angle clipping: Upper boundary , the lower boundary For each point in the extracted local point cloud data , calculate its angle with the horizontal plane :Based on point The three-dimensional coordinates of , calculate its angle with the XY plane . Judgment Point Angle Is it within the angle clipping boundary? If , then the point Mark the points that need to be retained, extract all the points that need to be retained, and obtain the local point cloud data related to the vehicle's driving slope. The selection of the preset angle range needs to consider factors such as the vehicle's climbing ability, terrain undulations, and the required map accuracy. The value is usually ±5° to ±15°. The local point cloud data after angle clipping is used as the pre-processed point cloud data for mining area map production.

[0066] According to the given content, formulate the mining map production standard including mining map element types and corresponding expression methods and usage scenarios, and determine the mining map element types to be expressed and the corresponding usage scenarios: Based on the obtained vehicle driving trajectory data, identify the road network in the mining area, and determine the road as an important map element type; Based on the obtained vehicle slope information, identify the slope change of the road, and determine the slope as another important map element type; Based on the obtained pre-processed point cloud data, identify the terrain undulations and distribution of objects around the road, and determine the terrain and objects as the third important map element type. Combined with the use requirements of unmanned vehicles in mining areas, determine the usage scenarios of each element type: road elements are mainly used for vehicle path planning and navigation, and need to express the geometric shape, width, length and other attributes of the road; slope elements are mainly used for vehicle speed control and braking, and need to express the longitudinal and transverse slope values ​​of the road; terrain and object elements are mainly used for vehicle obstacle avoidance and environmental perception, and need to express the terrain undulations and distribution of objects around the road.

[0067] Formulate road element expression standards, use vehicle driving trajectory data to extract the center line of the road: Smooth the vehicle driving trajectory to remove noise and abnormal points: Use the moving average filter method to smooth the vehicle driving trajectory data. Select an appropriate window size (such as 10 points), calculate the average coordinates of the points in the window, use the average coordinates as the new coordinates of the current point, and move the window in sequence to smooth all points on the trajectory. Detect and remove abnormal points on the smoothed trajectory. Calculate the distance and direction change between each point and its adjacent points before and after. If the distance or direction change exceeds the set threshold (such as the distance is greater than 10 meters or the direction change is greater than 90°), mark the point as an abnormal point and remove it from the trajectory. Use a curve fitting algorithm to fit the vehicle driving trajectory to obtain a smooth road center line: Select a spline curve algorithm (such as a cubic B-spline curve) to fit the smoothed vehicle driving trajectory. According to the number and distribution of trajectory points, determine the number of control points of the spline curve (such as setting a control point every 10 meters). Use the least squares method to calculate the coordinates of the control points of the spline curve to minimize the distance difference between the fitting curve and the trajectory points. The shape and smoothness of the fitting curve can be controlled by adjusting the node vectors of the spline curve (e.g., uniform distribution or adaptive distribution based on curvature). The fitted spline curve is discretized to generate a series of dense points (e.g., one point every 0.1 meters) as sampling points on the centerline of the road.

[0068] Taking the extracted road centerline as the skeleton, the preprocessed point cloud data is used to extract the road boundary on the left and right sides of the road centerline. With a certain width as the range, the boundary points in the preprocessed point cloud data are extracted: according to the road design standard or actual situation, an initial value of the road width (such as 6 meters) is set. Taking the extracted road centerline as the skeleton, half of the road width (such as 3 meters) is extended to the left and right sides to form a buffer zone. Traverse the preprocessed point cloud data and extract the point cloud points falling in the buffer zone as candidate points for the road boundary points. Downsample the extracted candidate points (such as voxelization or uniform sampling) to reduce the number of points and improve the efficiency of subsequent processing. Use the curve fitting algorithm to fit the left and right boundary points to obtain a smooth road boundary curve: divide the extracted candidate points into a left point set and a right point set according to the distance from the road centerline. Use the spline curve algorithm to fit the left point set and the right point set respectively to obtain the left and right boundary curves. According to the distribution of the boundary points, adjust the number of control points and node vectors of the spline curve to make the fitting curve better fit the actual road boundary. Discretize the left and right boundary curves to generate a series of dense points as sampling points on the boundary line. Combine the road centerline and boundary curves into a set of directed line segments to represent the geometric shape of the road: correspond the sampling points on the road centerline to the sampling points on the left and right boundary lines one by one to form a series of road cross sections. On each road cross section, take the centerline sampling point as the starting point, the left boundary sampling point as the second point, the right boundary sampling point as the third point, and the centerline sampling point as the end point to form a closed polyline to represent the geometric cross section of the road at that location. Connect the corresponding points of adjacent road cross sections in sequence to form a set of directed line segments to represent the three-dimensional geometric shape of the road.

[0069] Calculate the length of each line segment and mark it as the length attribute of the road: traverse each line segment that makes up the road geometry, calculate the three-dimensional space distance between the start and end points of the line segment, and add up the distances to get the total length of the road. Mark the total length of the road as the length attribute of the road, and mark the length of each line segment as the length attribute of the line segment. Calculate the distance between the left and right boundary curves and mark it as the width attribute of the road: on each road cross section, calculate the distance between the left and right boundary sampling points as the width of the road at that location. Traverse all road cross sections, calculate the minimum, maximum, and average values ​​of the road width, and mark them as the minimum width, maximum width, and average width attributes of the road, respectively. Mark the width of each road cross section as the width attribute of the road at that location.

[0070] Combine the centerline and boundary curves of the road into a set of directed line segments to represent the geometric shape of the road: For each sampling point on the centerline of the road, find the closest sampling point on the left and right boundary curves to form a road cross section: Traverse each sampling point on the centerline of the road, and for each sampling point, calculate its distance to all sampling points on the left and right boundary curves. Find the sampling point on the left boundary curve that is closest to the current centerline sampling point as the left boundary intersection point. Similarly, find the closest sampling point on the right boundary curve as the right boundary intersection point. Combine the current centerline sampling point, the left boundary intersection point, and the right boundary intersection point to form a road cross section. On the road cross section, take the centerline sampling point as the starting point and the left boundary sampling point as the end point to generate a directed line segment to represent the left half of the road: For each road cross section, take the centerline sampling point as the starting point and the left boundary intersection point as the end point, and connect to form a directed line segment.

[0071] Mark the directed line segment as the left half of the road, and define its direction as pointing from the start point to the end point. On the road cross section, take the centerline sampling point as the start point and the right boundary sampling point as the end point to generate a directed line segment to represent the right half of the road: for each road cross section, take the centerline sampling point as the start point and the right boundary intersection point as the end point, and connect them to form a directed line segment. Mark the directed line segment as the right half of the road, and define its direction as pointing from the start point to the end point. Connect the directed line segments of the left and right halves of the road end to end to form a closed polyline to represent the complete geometric shape of the road: traverse all sampling points on the center line of the road, and connect the left half directed line segments and the right half directed line segments corresponding to adjacent sampling points end to end in turn. At the start and end points, connect the end point of the last right half directed line segment with the start point of the first left half directed line segment to form a closed polyline. Use this closed polyline as the complete geometric shape of the road.

[0072] Calculate the length of each line segment and mark it as the length attribute of the road: For each directed line segment that makes up the road geometry, calculate the distance between its start and end points as the length of the line segment: Traverse each directed line segment that makes up the road geometry, and for each line segment, calculate the coordinate difference between its start and end points. Use the Pythagorean theorem to calculate the Euclidean distance between the start and end points and use it as the length of the line segment. Add up the lengths of each line segment to get the total length of the road and mark it as the length attribute of the road: Traverse all directed line segments that make up the road geometry and add up the lengths of each line segment to get the total length of the road. Mark the total length of the road as the length attribute of the road.

[0073] Calculate the distance between the left and right boundary curves and mark it as the width attribute of the road: For each sampling point on the center line of the road, calculate its distance to the corresponding sampling point on the left and right boundary curves as the width of the road at that location: Traverse each sampling point on the center line of the road, and for each sampling point, calculate the coordinate difference between it and the intersection of the left and right boundaries. Use the Pythagorean theorem to calculate the Euclidean distance between the center line sampling point and the intersection of the left and right boundaries, and use it as the width of the road at that location. Take the average of the widths at various locations on the road as the average width of the road and mark it as the width attribute of the road: Calculate the arithmetic mean of the widths at various locations on the road and use it as the average width of the road. Mark the average width of the road as the width attribute of the road. Take the minimum and maximum values ​​of the widths at various locations on the road as the minimum and maximum widths of the road and mark them as the width range attribute of the road: Find the minimum and maximum values ​​among the widths at various locations on the road and use them as the minimum and maximum widths of the road, respectively. The minimum and maximum widths of the road are combined into an interval and marked as the width range attribute of the road.

[0074] Extract mining map elements from preprocessed point cloud data, classify preprocessed point cloud data using deep learning-based point cloud semantic segmentation method, and extract point cloud subsets of different categories including roads, stop lines, loading areas, unloading areas, and parking lots: Select appropriate point cloud semantic segmentation deep learning models, such as PointNet++, SPGraph, etc., and fine-tune and train the models according to the characteristics of the mining environment and the categories of elements to be extracted. Input the preprocessed point cloud data into the trained semantic segmentation model, classify each point, and obtain point cloud subsets belonging to roads, stop lines, loading areas, unloading areas, and parking lots. For each category of point cloud subset, perform post-processing operations, such as removing outliers and smoothing point clouds, to improve the quality of point clouds.

[0075] Perform Euler clustering on the extracted road point cloud subset to obtain the road centerline point cloud, and use the least squares method to fit the road centerline point cloud to extract the lane centerline element: Perform Euler clustering on the road point cloud subset to aggregate points with close spatial distance into a cluster, and each cluster represents a road. For each road cluster, the road centerline point cloud is obtained by calculating the centroid or central axis of the point cloud. Use the least squares method to fit the road centerline point cloud to obtain a smooth curve as the lane centerline element. Polynomial fitting, spline curve fitting and other methods can be used. Simplify and smooth the lane centerline obtained by fitting to reduce redundant points and improve the quality of the centerline. Perform plane fitting on the point cloud subset of the loading area, unloading area and parking lot functional areas, extract the plane boundary line, and obtain the vector representation of the elements of each functional area: Perform plane fitting on the point cloud subset of the loading area, unloading area and parking lot, and estimate the plane equation through algorithms such as RANSAC. According to the plane equation obtained by fitting, calculate the boundary line of the plane, and use the convex hull algorithm or alpha shape algorithm to extract the boundary line. The extracted boundary lines are simplified and smoothed to reduce redundant points and obtain vector representation of functional area elements. Attribute information such as area type and area is added to each functional area element.

[0076] Extract feature points from point cloud data using a key point detection-based method and use the feature points as POI elements: Select a suitable key point detection algorithm, such as ISS, SIFT-3D, etc., and set the algorithm parameters according to the characteristics of the mining environment and the type of POI to be extracted. Perform key point detection on the preprocessed point cloud data to extract points with significant features, such as sharp points, corner points, etc. Filter and cluster the extracted key points to remove redundant points and outliers to obtain POI elements. Add attribute information to each POI element, such as POI type, coordinates, etc. Construct a topological connectivity graph based on the extracted lane lines, and determine the road right relationship between lanes by analyzing the topological connectivity graph: Topologically construct the extracted lane centerline elements to generate a topological connectivity graph. Graph theory algorithms, such as minimum spanning tree and shortest path, can be used. In the topological connectivity graph, the lane centerline elements are used as edges, and the endpoints and intersections of the lane centerlines are used as nodes. By analyzing the structure and attributes of the topological connectivity graph, such as node degree, edge direction, etc., the right-of-way relationship between lanes is determined, such as main roads, branch roads, intersections, etc. Right-of-way attributes are added to each lane centerline element, such as road grade, travel direction, etc.

[0077] Cartographic tool development, the extracted mining map elements are classified according to the geometric and attribute characteristics of points, lines, surfaces and relationships, and the geometric expression and attribute information of each type of mining map elements are set according to the set mining map production standards: the extracted mining map elements are classified, and the points of interest POI are divided into point types, the center line of the lane and the stop line are divided into line types, the loading area, unloading area and parking lot are divided into surface types, and the right of way relationship is divided into relationship types. According to the mining map production standards, the geometric expression method is set for each type of element, such as point elements are represented by icons or symbols, line elements are represented by polylines or curves, surface elements are represented by polygons or surfaces, and relationship elements are represented by directed line segments. Attribute information is set for each type of element, such as the name and type of point elements, the width, length, and driving direction of line elements, the material type, number of parking spaces, and entrance and exit locations of surface elements, and the direction of travel, turn type, priority, etc. of relationship elements.

[0078] Pre-develop general functional modules such as drawing, editing, rendering, and management of basic map elements (points, lines, surfaces, and relationships): Develop point element drawing functions, support adding, deleting, and moving point elements on the map, and set the symbol style and size of point elements. Develop line element drawing functions, support drawing polylines or curves on the map, and set the color, width, and line type of line elements. Develop surface element drawing functions, support drawing polygons or surfaces on the map, and set the fill color and boundary style of surface elements. Develop relationship element drawing functions, support drawing directed line segments on the map, and set the arrow style and direction of relationship elements. Develop feature editing functions, support modifying, deleting, copying, and other operations on drawn features, and support setting feature attribute information. Develop feature rendering functions, support symbolizing, annotating, and stylizing features according to feature attribute information and set rendering rules. Develop feature management functions, support layer management of drawn features, and support operations such as adding, deleting, hiding, showing, and sorting layers. For point-type mining map elements, copy the geometric drawing and attribute editing functions of the point elements, and build a POI drawing module: copy the drawing and editing functions of the point elements in the general function module, and build a POI drawing module. In the POI drawing module, receive the location information of the POI, call the point element drawing function, and draw the geometric representation of the POI at the specified location. In the POI drawing module, receive the attribute information of the POI, such as name, type, etc., call the attribute editing function, and associate the attribute information with the drawn point element.

[0079] For line-type mining map elements, copy the geometric drawing and attribute editing functions of the line elements, and build the lane centerline and stop line drawing modules: Copy the drawing and editing functions of the line elements in the general function module, and build the lane centerline and stop line drawing modules. In the lane centerline drawing module, receive the position and direction information of the lane centerline, call the line element drawing function, and draw the geometric representation of the lane centerline according to the specified position and direction. In the lane centerline drawing module, receive the attribute information of the lane centerline, such as width, length, driving direction, etc., call the attribute editing function, and associate the attribute information with the drawn line element. In the stop line drawing module, receive the position information of the stop line, call the line element drawing function, and draw the geometric representation of the stop line at the specified position. In the stop line drawing module, receive the attribute information of the stop line, such as type, etc., call the attribute editing function, and associate the attribute information with the drawn line element.

[0080] For the surface type mining map elements, copy the geometric drawing and attribute editing functions of the surface elements, and build the loading area, unloading area and parking lot drawing modules: copy the surface element drawing and editing functions in the general function module, and build the loading area, unloading area and parking lot drawing modules. In the loading area and unloading area drawing modules, receive the boundary line information of the loading area and unloading area, call the surface element drawing function, and draw the geometric representation of the loading area and unloading area according to the specified boundary line. In the loading area and unloading area drawing modules, receive the attribute information of the loading area and unloading area, such as material type, call the attribute editing function, and associate the attribute information with the drawn surface elements. In the parking lot drawing module, receive the boundary line information of the parking lot, call the surface element drawing function, and draw the geometric representation of the parking lot according to the specified boundary line. In the parking lot drawing module, receive the attribute information of the parking lot, such as the number of parking spaces, entrance and exit locations, etc., call the attribute editing function, and associate the attribute information with the drawn surface elements.

[0081] For the mine map elements of the road right relationship type, copy the attribute editing function of the relationship element and build a road right relationship drawing module: copy the attribute editing function of the relationship element in the general function module and build a road right relationship drawing module. In the road right relationship drawing module, the road right relationship is geometrically expressed through directed line segments, the starting and ending position information of the road right relationship is received, and the relationship element drawing function is called to draw the directed line segments representing the road right relationship. In the road right relationship drawing module, the attribute information of the road right relationship, such as the direction of travel, the turn type, the priority, etc., is received, and the attribute editing function is called to associate the attribute information with the drawn relationship line segments. Transplant the rendering display and feature layer management functions to each mine map element drawing module to realize the visual display and layer management of the mine map elements: transplant the rendering display function in the general function module to each mine map element drawing module, and symbolize, annotate and style the drawn mine map elements according to the attribute information of the elements and the set rendering rules to realize the visual display of the mine map elements. The feature layer management function in the general function module is transplanted to each mining area map feature drawing module, which supports layer management of the drawn mining area map elements and realizes operations such as adding, deleting, hiding, displaying and sorting layers, making it convenient for users to organize and manage mining area map elements.

[0082] Mine map production: Integrate the constructed mine map element drawing module into the mapping tool; for different mine projects, if the definition of the same mine map element is different, set the label of the corresponding mine map element drawing module according to the project, and control the access rights of different project users to the mine map element drawing module through user authority management; use the integrated mapping tool to create or select the project label corresponding to the mine map to be produced; under the selected project label, call the corresponding mine map element drawing module from the mapping tool interface according to the needs of the mine map; specifically, if it is necessary to draw a point of interest POI, call the point of interest POI drawing module; if it is necessary to draw a lane centerline or stop line, call the lane centerline drawing module or the stop line drawing module; if it is necessary to draw a loading area, unloading area or parking lot, call the loading area drawing module, unloading area drawing module or parking lot drawing module; if it is necessary to edit the road right relationship, call the road right relationship drawing module. In the called mining area map element drawing module, the geometric shapes of mining area map elements are drawn on the map by means of human-computer interaction using a mouse or other input devices; after the drawing is completed, the attribute information of the drawn mining area map elements is entered into the drawing module, including the name, type, description, etc.; the drawn mining area map elements are rendered and displayed, and added to the map in the form of layers, and the display and hiding of the feature layers can be controlled by the feature layer management function in the drawing module; the execution is repeated until all required mining area map elements are drawn; and a complete mining area map is automatically generated based on the drawn mining area map elements.

Claims

1. A method for making a mining area map based on a user-defined mapping tool, comprising: Obtain mining area terrain and vehicle specification parameters, and generate mining area road network according to the mining area terrain and vehicle specification parameters; The inertial measurement unit (IMU) and the dynamic positioning system (RTK) are used to collect the vehicle's driving trajectory and slope information on the mining area road network, and the collected slope information is converted into the slope information when the vehicle is in the forward driving state. Forward driving refers to the driving state in which the vehicle's forward direction is consistent with the lane direction. Use vehicle-mounted laser radar to collect 3D point cloud data of the mining area, and perform denoising and cropping preprocessing on the collected point cloud data; Based on the vehicle driving trajectory data, converted slope information and pre-processed point cloud data, formulate the mining area map production standards including mining area map element types and corresponding expressions and usage scenarios; According to the set mining area map production standards, mining area map elements are extracted from the pre-processed point cloud data; mining area map elements include points of interest (POI), lane center lines, stop lines, loading areas, unloading areas, parking lots and road rights; among them, road rights refer to the passage and locking relationship between lanes; The extracted mining map elements are classified by geometric and attribute characteristics according to points, lines, surfaces and relationships, and the geometric expression and attribute information of each type of mining map element are set according to the set mining map production standards; Pre-set the basic map element drawing function of points, lines, surfaces and relationships in the mapping tool; based on the classified mining area map elements, copy and transplant the basic map element drawing function through computer programming to build a mining area map element drawing module that is compatible with the geometric expression and attribute information of each type of mining area map element; wherein the basic map element drawing function includes geometric drawing, geometric editing, attribute editing, rendering display and element layer management; Integrate the constructed mining area map element drawing module into the mapping tool; for different mining area projects, if the definitions of the same mining area map elements are different, set labels for the corresponding mining area map element drawing modules according to the projects, and control the access rights of different project users to the mining area map element drawing modules through user authority management; Utilize the integrated mapping tools and call the mining map element drawing module under the corresponding project tag according to the mining map requirements of the project, draw the mining map elements, and generate the mining map.

2. The method for making a mining area map based on a user-defined mapping tool according to claim 1, characterized in that: Generate a mine road network based on the mine terrain and vehicle specification parameters, including: Acquire the topographic data of the mining area, which includes the geometric parameters of the mining area roads and the distribution information of the mining area objects; Obtain vehicle specifications in the mining area, including vehicle size, load capacity and power performance; Constructing a three-dimensional terrain model of the mining area based on the acquired mining area terrain data, and identifying graphic elements including roads and landforms in the three-dimensional terrain model of the mining area; Based on the acquired mining area vehicle specification parameters, the vehicle's motion characteristics are simulated using vehicle dynamics simulation in the mining area's three-dimensional terrain model to obtain the vehicle's driving ability parameters on the mining area's roads; The width, slope and turning radius of the mining area roads are constrained according to the driving capacity parameters to generate a mining area road network model that meets the vehicle traffic requirements.

3. The method for making a mining area map based on a user-defined mapping tool according to claim 2, characterized in that: The inertial measurement unit (IMU) and the dynamic positioning system (RTK) are used to collect the vehicle's driving trajectory and slope information on the mining area road network, including: The inertial measurement unit (IMU) is used to collect the vehicle's attitude information on the mining area road network, and the dynamic positioning system (RTK) is used to collect the vehicle's position information on the mining area road network. Based on the collected attitude information and position information, the vehicle's driving trajectory on the mining area road network is calculated; the attitude information includes the vehicle's yaw angle, pitch angle, and roll angle, and the position information includes the vehicle's latitude and longitude coordinates and elevation; The inertial measurement unit (IMU) is used to collect the acceleration information of the vehicle during driving. Based on the collected acceleration information, the acceleration vector decomposition method is used to calculate the slope information of the vehicle relative to the horizontal plane. The vehicle's heading angle information and lane direction information are obtained, and the angle between the vehicle's heading angle and the lane direction is calculated. When the absolute value of the angle is less than or equal to a preset threshold, the vehicle is judged to be in a forward driving state, and the calculated slope information remains unchanged; when the absolute value of the angle is greater than the preset threshold, the vehicle is judged to be in a non-forward driving state, and the calculated slope information is multiplied by -1 to obtain the corresponding slope information when the vehicle is in a forward driving state; wherein, the vehicle's heading angle information is obtained through a vehicle-mounted inertial navigation system or a vehicle-mounted electronic compass, and the lane direction information is extracted from the calculated vehicle driving trajectory.

4. The method for making a mining area map based on a user-defined mapping tool according to claim 3, characterized in that: Based on the collected posture information and position information, the vehicle's driving trajectory on the mining area road network is calculated, including: The inertial measurement unit (IMU) is used to continuously collect the vehicle's angular velocity and acceleration information at a preset sampling frequency, and the dynamic positioning system (RTK) is used to continuously collect the vehicle's position information at a preset sampling frequency, and the timing information of the vehicle's attitude and position is obtained synchronously; the sampling frequency of the IMU is higher than the sampling frequency of the RTK, and the information collected by the two sensors is aligned through timestamps; The collected angular velocity and acceleration information are corrected and aligned using the zero-speed correction algorithm and the static alignment algorithm to eliminate the zero bias error and installation error of the IMU and obtain the corrected angular velocity and acceleration information; Using the Kalman filter algorithm, the corrected angular velocity and acceleration information is fused with the position information collected by RTK to estimate the vehicle's attitude and position state, and obtain the vehicle's attitude information and position information; According to the posture information of the vehicle, the speed information in the vehicle coordinate system is converted to the mine road network coordinate system using the Euler angle conversion formula to obtain the speed information of the vehicle in the mine road network coordinate system; According to the location information of the vehicle, the longitude and latitude coordinates and elevation of the vehicle are converted into the mine road network coordinate system using the coordinate conversion formula to obtain the location information of the vehicle in the mine road network coordinate system; The Runge-Kutta integration algorithm is used to integrate the obtained speed information with respect to time and superimpose it with the obtained position information to obtain the vehicle's driving trajectory in the mining area road network coordinate system.

5. The method for making a mining area map based on a user-defined mapping tool according to claim 4, characterized in that: Get the vehicle's driving trajectory in the mining area road network coordinate system, including: The speed information of the vehicle in the mining area road network coordinate system is recorded as , the location information is recorded as ; According to the Runge-Kutta integration algorithm, the time t is discretized into , the time step is h, then ; and speed and location In time The values ​​at and ; Using the fourth-order Runge-Kutta integral formula, calculate the speed The integral increment within the time step h : ； ； ； ； ； in, Indicates at time ,Location The speed value at Leverage Location With speed integral increment , iteratively calculate the position at the next moment : ； Repeat the iteration until the final moment is calculated Location , obtained by to The vehicle driving trajectory.

6. The method for making a mining area map based on a user-defined mapping tool according to claim 5, characterized in that: ； in, For the initial moment location, Indicates the time interval Internal speed Perform a definite integral.

7. The method for making a mining area map based on a user-defined mapping tool according to claim 4, characterized in that: The collected point cloud data is pre-processed by denoising and cropping, including: The vehicle-mounted laser radar is used to collect the three-dimensional point cloud data of the mining area at a preset scanning frequency and scanning angle to obtain the original point cloud data reflecting the topographical features of the mining area; De-noising the original point cloud data: Calculate the density of each point in the original point cloud data, mark the points with density less than the threshold as noise points, and use the statistical filtering method to remove the noise points to obtain point cloud data with uniform density; The octree algorithm is used to voxelize the point cloud data with uniform density, merge the points in each voxel, and replace the original points with the voxel center point to obtain the point cloud data with reduced density; The curvature of each point in the point cloud data with reduced density is calculated using the Laplace operator, and the points with curvature greater than the threshold are subjected to Laplace smoothing to obtain point cloud data with smooth surface as the denoised point cloud data; The denoised point cloud data is cropped to obtain preprocessed point cloud data.

8. The method for making a mining area map based on a user-defined mapping tool according to claim 7, characterized in that: The denoised point cloud data is cropped, including: The denoised point cloud data and the obtained vehicle driving trajectory are spatially calibrated, and a preset spatial range is set with the vehicle driving trajectory as the center. The point cloud data is spatially cropped using the spatial range to extract local point cloud data related to the vehicle driving path; Based on the slope information obtained when the vehicle is driving forward, a preset angle range is set, and the extracted local point cloud data is clipped using the angle range to obtain local point cloud data related to the vehicle's driving slope, which is used as preprocessed point cloud data for mining area map production.

9. The method for making a mining area map based on a user-defined mapping tool according to claim 7, characterized in that: Develop mining area map production standards that include mining area map element types and corresponding expressions and usage scenarios, including: Determine the types of mining area map elements to be expressed, including roads, slopes, and terrain features, based on the obtained vehicle driving trajectory data, the obtained vehicle slope information, and the obtained pre-processed point cloud data, and determine the usage scenarios of each element type based on the usage requirements of the mining area unmanned vehicles; According to the type of each element, formulate the corresponding element expression standard: For road features, the road centerline is extracted using the vehicle driving trajectory, and the road centerline is used as the skeleton to extract the road boundary using the pre-processed point cloud data, expressed as a set of directed line segments, and the road width and length attributes are annotated; For slope elements, the longitudinal and transverse slope values ​​of the road are calculated using the vehicle slope information and pre-processed point cloud data, and the slope values ​​are divided into different levels. Different colors or symbols are used to represent different levels of slopes and superimposed on the road elements; For terrain and object elements, the pre-processed point cloud data is used to extract the terrain undulation information and object distribution information within the preset range of the road, and expressed as different geometric primitives according to the characteristics of the terrain and objects; The set feature expression standards are associated with the determined feature types and corresponding usage scenarios as the standards for mining area map production.

10. The method for making a mining area map based on a user-defined mapping tool according to claim 9, characterized in that: Extract mining area map elements, including: According to the established mining area map production standards, mining area map elements are extracted from the obtained pre-processed point cloud data: The pre-processed point cloud data is classified using a point cloud semantic segmentation method based on deep learning, and point cloud data of different categories such as roads, stop lines, loading areas, unloading areas, and parking lots are extracted; Perform Euler clustering on the extracted road point cloud subset to obtain the road centerline point cloud, and use the least squares method to fit the road centerline point cloud to extract the lane centerline elements; Perform plane fitting on the point cloud subsets of the loading area, unloading area and parking lot functional areas, extract the plane boundary lines, and obtain the vector representation of the elements of each functional area; use the key point detection method to extract feature points from the point cloud data, and use the feature points as point of interest (POI) elements; Based on the extracted lane lines, a topological connectivity graph is constructed, and the road right relationship between lanes is determined by analyzing the topological connectivity graph.

Images