Strip mine map construction and updating method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention relates to the technical field of automatic driving, and discloses a strip mine map construction and updating method and device, electronic equipment and a storage medium, and the method comprises the steps: constructing a strip mine map comprising a static map, a road driving area map, a loading area map, a dumping site unloading area map and an obstacle map, performing real-time map updating on the road driving area map, the loading area map, the dumping site unloading area map and the obstacle map; wherein the dumping site unloading area map comprises a retaining wall curve, the retaining wall curve is obtained by processing a comprehensive data matrix through a 3D artificial intelligence model, the comprehensive data matrix is obtained by combining a plurality of data matrixes, and each data matrix comprises coarse curves of a plurality of retaining walls; the coarse curve of each retaining wall is determined based on a frame of sensor data collected during the movement of the driverless truck. According to the mode, a real-time strip mine high-precision map is provided so as to guide, plan and decide operation of various mining area automatic driving vehicles.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a method, apparatus, electronic device and storage medium for constructing and updating open-pit mine maps. Background Technology

[0002] With the continuous development of autonomous driving technology, driverless mining trucks, mining trucks, bulldozers, and other vehicles have become important components of intelligent mining solutions. To facilitate the operation of driverless vehicles in mines, it is necessary to construct a real-time high-precision map of open-pit mines to guide, plan, and make decisions regarding the operation of various vehicles, thereby improving operational efficiency and preventing safety accidents. Summary of the Invention

[0003] The purpose of this application is to provide at least one method, apparatus, electronic device and storage medium for constructing and updating open-pit mine maps, which can provide at least one real-time high-precision map of open-pit mines to facilitate the guidance, planning and decision-making of the operation of various autonomous vehicles in mining areas.

[0004] To address the aforementioned technical problems, at least one embodiment of this application provides a method for constructing and updating an open-pit mine map, comprising: constructing an open-pit mine map including a static map, a road driving area map, a loading area map, a spoil heap unloading area map, and an obstacle map; and performing real-time map updates on the road driving area map, the loading area map, the spoil heap unloading area map, and the obstacle map; wherein, the spoil heap unloading area map includes retaining wall curves, the retaining wall curves being obtained by processing a comprehensive data matrix using a 3D artificial intelligence model, the comprehensive data matrix being obtained by merging multiple data matrices, each data matrix including multiple coarse curves of retaining walls, each coarse curve of a retaining wall being determined based on a frame of sensor data collected during the movement of an unmanned truck, the movement process being the process by which the unmanned truck moves from entering the spoil heap unloading area to moving to the corresponding collision critical point, the sensor data including image data and laser point cloud data, and one data matrix corresponding to one unmanned truck.

[0005] By dividing the open-pit mine map into static maps, road driving area maps, loading area maps, spoil heap unloading area maps, and obstacle maps, it is easier to adopt corresponding map construction and update methods for different areas, which is beneficial for building high-precision maps. Furthermore, the spoil heap unloading area focuses on determining the precise location of retaining walls. Sensor data collected by unmanned trucks during their movement in the spoil heap unloading area is used to construct a coarse curve for the retaining wall, forming a data matrix corresponding to the unmanned truck. Multiple data matrices are then used to construct a comprehensive data matrix, which is further processed using a 3D artificial intelligence model to generate the final retaining wall curve. This method results in a high-precision retaining wall curve, which in turn facilitates the construction of a high-precision open-pit mine map. This allows various mining trucks to operate according to the map, improving operational efficiency and preventing safety accidents.

[0006] Furthermore, by using the unmanned trucks currently in operation as a mobile sensor network to continuously and collaboratively perceive the environment, not only is the real-time performance and efficiency of map updates for the spoil heap unloading area improved, ensuring that the unmanned trucks always make decisions based on the latest environmental information, but more importantly, it enables the newly opened spoil heap unloading area to quickly build a usable high-precision map "from scratch".

[0007] In some optional embodiments, the update process of the spoil heap unloading area map includes: receiving sensor data sent by unmanned trucks operating in the spoil heap unloading area in real time, the sensor data including first pose data and first laser point cloud data; constructing a three-dimensional point cloud model of the spoil heap unloading area based on the sensor data; extracting the unloading platform edges, leveled areas, and unloading points in the spoil heap unloading area from the three-dimensional point cloud model; and updating the spoil heap unloading area map based on the unloading platform edges, leveled areas, and unloading points in the spoil heap unloading area, wherein the spoil heap unloading area map also marks the status of each unloading point, the status of the unloading point being one of idle, occupied, and reserved.

[0008] By constructing a 3D point cloud model and extracting the edges of the unloading platform, leveled areas, and unloading points, the complex and irregular terrain of the spoil heap is transformed into structured data that computers can understand and process. Labeling each unloading point with its "idle," "occupied," or "reserved" status allows for clear visibility of which unloading points are in operation, available, or planned when dispatching vehicles. This enables precise and efficient task allocation and route planning for multiple trucks, preventing congestion or waiting caused by vehicles converging on the same area and preventing vehicles from accidentally entering dangerous areas.

[0009] In some optional embodiments, the process of updating the loading area map includes: receiving sensor data from electric loaders and / or bulldozers operating in the loading area in real time, the sensor data including second pose data and second laser point cloud data; constructing a three-dimensional point cloud model of the loading area based on the sensor data; segmenting the three-dimensional point cloud model of the loading area into a ground point cloud model, a working face point cloud model, a ore pile point cloud model, an equipment point cloud model, and a miscellaneous point cloud model, and extracting the boundary line of the working face, and determining the drivable area in the ground corresponding to the ground point cloud model; and updating the loading area map according to the ore pile point cloud model, the equipment point cloud model, the miscellaneous point cloud model, the boundary line of the working face, and the drivable area.

[0010] By segmenting the 3D point cloud model into different categories such as ground, working face, ore pile, and equipment, it is easy to accurately distinguish where the vehicle can travel (drivable area) and where the loading points are (ore pile and working face). The extracted working face boundary lines and the determined drivable areas can facilitate the planning of safe and efficient approach, parking, and departure paths for unmanned trucks, and help solve the problem of collision avoidance when multiple large pieces of equipment interact closely in narrow and dynamic loading areas.

[0011] In some optional embodiments, the process of updating the road driving area map includes: receiving point cloud maps and vehicle trajectories sent by a data collection vehicle operating in the road driving area in real time; determining lane line markings on the road corresponding to the vehicle trajectory in the point cloud map based on point cloud intensity and shape recognition; extracting the boundary lines of the road corresponding to the vehicle trajectory and determining the road centerline based on the boundary lines; updating the road topology network of the road driving area according to the road corresponding to the vehicle trajectory; and updating the road driving area map according to the updated road topology network, the lane line markings, the boundary lines, and the road centerline.

[0012] Mining area roads experience frequent changes in condition (such as lane markings and shoulders) due to heavy vehicle traffic and rain erosion. By continuously collecting data via data collection vehicles and automatically identifying lane markings, extracting road boundaries and centerlines, and updating the road topology network, the accuracy of road information relied upon for navigation and route planning is ensured. The updated road topology network clearly defines road connections, driving directions, and rules (such as intersection priorities), facilitating more scientific and rational vehicle scheduling and route planning. Simultaneously, precise lane marking and road boundary information provides clear tracking references for autonomous trucks, enhancing their stability and safety in complex road conditions. Continuous maintenance of the road network on the map is a fundamental prerequisite for ensuring the efficient and orderly operation of the entire mining logistics system, preventing route planning errors, traffic congestion, and even safety accidents caused by outdated maps.

[0013] In some optional embodiments, the obstacle map base update process includes: receiving obstacle information sent in real time by vehicles and / or drones traveling in the open-pit mine corresponding to the open-pit mine map, the obstacle information including the type, location, size and sensor data of the obstacle; displaying the obstacle information to the verifier and receiving the verification result fed back by the verifier; when the verification result confirms the obstacle, updating the obstacle map according to the type, location and size of the obstacle.

[0014] Temporary obstacles in the mining environment (such as falling rocks, accumulated water, and temporarily parked vehicles) pose a major threat to driving safety. By implementing a process of "vehicle perception and reporting - manual verification and confirmation - map update and distribution," each moving truck becomes a mobile "sentinel," forming a dynamic safety monitoring network covering the entire mining area. The introduction of manual verification effectively utilizes human advantages in image and contextual understanding, filtering out false alarms from sensors (such as shadows) and updating only real and persistent threats to the map. This avoids frequent false alarms or unnecessary detours caused by data noise, ensuring the authority and reliability of obstacle information.

[0015] In some optional embodiments, the method further includes: after any of the road driving area map, loading area map, spoil heap unloading area map, and obstacle map is updated, sending an update notification to the vehicle operating in the open-pit mine, so that the vehicle can download the latest version of any of the maps indicated in the update notification; wherein, the vehicle is configured to initiate a self-check, the self-check being: in response to its own vehicle startup, requesting the latest version number of each of the road driving area map, loading area map, spoil heap unloading area map, and obstacle map from the cloud, determining the local map that needs to be updated based on the latest version number, and downloading the latest version of the local map from the cloud; the vehicle is also configured to periodically perform the self-check according to a preset check duration.

[0016] In the complex communication environment of mining areas, network outages and frequent vehicle starts and stops are common occurrences. A multi-layered, complementary, and reliable update strategy is formed by combining three mechanisms: "proactive push," "initiation self-check," and "periodic checks." Proactive push ensures that map updates are distributed to vehicles with minimal latency when network conditions are good. Initiation self-check ensures that vehicles are synchronized with the latest map before each trip, avoiding the risk of carrying outdated maps. Periodic checks serve as a backup for the first two mechanisms, compensating for missed push messages or maintaining synchronization during long-term operation. This design fully considers the actual challenges of industrial sites, ensuring that multiple vehicles use the latest and consistent map versions under complex conditions. This is the foundation for achieving multi-vehicle collaborative scheduling and avoiding planning conflicts caused by map version differences (such as two vehicles being planned to the same location based on different maps).

[0017] At least one embodiment of this application also provides an open-pit mine map construction and updating device, including a construction module for constructing an open-pit mine map comprising a static map, a road driving area map, a loading area map, a spoil heap unloading area map, and an obstacle map; and an updating module for real-time map updates of the road driving area map, the loading area map, the spoil heap unloading area map, and the obstacle map; wherein, the spoil heap unloading area map includes retaining wall curves, the retaining wall curves being obtained by processing a comprehensive data matrix using a 3D artificial intelligence model, the comprehensive data matrix being obtained by merging multiple data matrices, each data matrix including multiple coarse curves of retaining walls, each coarse curve of a retaining wall being determined based on a frame of sensor data collected during the movement of an unmanned truck, the movement being the process by which the unmanned truck moves from entering the spoil heap unloading area to moving to the corresponding collision critical point, the sensor data including image data and laser point cloud data, and one data matrix corresponding to one unmanned truck.

[0018] In some optional embodiments, the updating module is configured to receive sensor data sent in real time from unmanned trucks operating in the spoil heap unloading area, the sensor data including first pose data and first laser point cloud data; construct a three-dimensional point cloud model of the spoil heap unloading area based on the sensor data; extract the unloading platform edges, leveled areas, and unloading points in the spoil heap unloading area from the three-dimensional point cloud model; and update the spoil heap unloading area map based on the unloading platform edges, leveled areas, and unloading points in the spoil heap unloading area, wherein the spoil heap unloading area map also marks the status of each unloading point, the status of the unloading point being one of idle, occupied, and reserved.

[0019] In some optional embodiments, the update module is configured to receive sensor data from electric loaders and / or bulldozers operating in the loading area in real time, the sensor data including second pose data and second laser point cloud data; construct a three-dimensional point cloud model of the loading area based on the sensor data; segment the three-dimensional point cloud model of the loading area into a ground point cloud model, a working face point cloud model, a ore pile point cloud model, an equipment point cloud model, and a miscellaneous point cloud model, and extract the boundary line of the working face, and determine the drivable area in the ground corresponding to the ground point cloud model; update the loading area map according to the ore pile point cloud model, the equipment point cloud model, the miscellaneous point cloud model, the boundary line of the working face, and the drivable area.

[0020] In some optional embodiments, the updating module is configured to receive point cloud maps and vehicle trajectories sent in real time by a data collection vehicle operating in the road driving area; determine lane line markings on the road corresponding to the vehicle trajectory in the point cloud map based on point cloud intensity and shape recognition; extract the boundary lines of the road corresponding to the vehicle trajectory, and determine the road centerline based on the boundary lines; update the road topology network of the road driving area according to the road corresponding to the vehicle trajectory; and update the road driving area map according to the updated road topology network, the lane line markings, the boundary lines, and the road centerline.

[0021] In some optional embodiments, the update module is configured to receive obstacle information sent in real time by vehicles and / or drones traveling in the open-pit mine corresponding to the open-pit mine map, the obstacle information including the type, location, size and sensor data of the obstacle; display the obstacle information to the verifier and receive the verification result fed back by the verifier; when the verification result confirms the obstacle, update the obstacle map according to the type, location and size of the obstacle.

[0022] In some optional embodiments, the apparatus further includes a synchronization module for sending an update notification to vehicles operating in the open-pit mine after any of the road driving area map, loading area map, spoil heap unloading area map, and obstacle map has been updated, so that the vehicles can download the latest version of any of the maps indicated in the update notification.

[0023] At least one embodiment of this application also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-described open-pit mine map construction and updating method.

[0024] At least one embodiment of this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for constructing and updating open-pit mine maps.

[0025] At least one embodiment of this application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for constructing and updating open-pit mine maps. Attached Figure Description

[0026] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.

[0027] Figure 1This is a flowchart of an embodiment of the method for constructing and updating open-pit mine maps provided in this application; Figure 2 This is a flowchart of updating a spoil heap unloading area map provided in one embodiment of this application; Figure 3 This is a flowchart of an update process for a loading area map provided in one embodiment of this application; Figure 4 This is a flowchart of a road driving area map update provided in one embodiment of this application; Figure 5 This is a flowchart of an obstacle map update provided in one embodiment of this application; Figure 6 This is a schematic diagram of an open-pit mine map construction and updating device provided in another embodiment of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0029] Example 1: The open-pit mine map construction and updating method of this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. Its specific process can be as follows: Figure 1 As shown, it includes S101-S102.

[0030] S101, Construct an open-pit mine map that includes a static map, a road driving area map, a loading area map, a spoil heap unloading area map, and an obstacle map.

[0031] The initially constructed road driving area map, loading area map, spoil heap unloading area map, and obstacle map can be empty. That is, the contents of the corresponding map files are empty. When the road driving area map, loading area map, spoil heap unloading area map, and obstacle map are updated later, the corresponding maps will be stored in the corresponding map files.

[0032] Static maps are maps of static areas in open-pit mines that generally do not change, such as office areas and gas stations. In some examples, the process of constructing a static map may include: acquiring images of the static area taken by drones or satellites, using these images as a base map, and manually drawing polygons or points such as office areas, gas stations, parking lots, and restricted areas on the map to obtain a static map.

[0033] S102 updates the road driving area map, loading area map, spoil heap unloading area map, and obstacle map in real time.

[0034] In some examples, the unloading area map of the spoil heap includes retaining wall curves, which are obtained by processing a comprehensive data matrix using a 3D artificial intelligence model. The comprehensive data matrix is ​​obtained by merging multiple data matrices, each of which includes coarse curves of multiple retaining walls. The coarse curve of each retaining wall is determined based on a frame of sensor data collected during the movement of the autonomous truck. The movement process is the process from the autonomous truck entering the spoil heap unloading area to moving to the corresponding collision critical point. The sensor data includes image data and laser point cloud data. One data matrix corresponds to one autonomous truck.

[0035] For example, when the first vehicle M1 (an autonomous truck, not described further) enters the spoil heap unloading area, its position P1 is recorded. Based on the onboard camera data C, the laser point cloud data L, and their calibration parameters T, the coarse curve A1B1 of the retaining wall can be obtained. The spoil heap area L, which has the lowest curvature and is closest to the vehicle, is selected, where L is not less than twice the width of the vehicle (considering that there are often two or more vehicles entering the spoil heap unloading area sequentially or simultaneously for spoil heap operations, and that the first acquisition is the coarse curve of the retaining wall, the vehicle's attitude needs to be further adjusted before the final spoil heap stopping point is determined, therefore a sufficiently wide area is required). Then, the vehicle determines its initial running trajectory based on the coarse curve and the normal F of the selected area L, and automatically runs to the collision threshold P2 (P2 is determined based on the selected area L) in a collision-free area.

[0036] During operation, as the vehicle moves from P1 to P2, a series of coarse curve data A1B1~AnBn are generated based on sensor data, denoted as matrix [P, AB] (data matrix). Simultaneously, other vehicles gradually enter the spoil heap unloading area. The coarse curve of the retaining wall obtained based on the sensor data of vehicles M2~Mn, along with the coarse curve data of the retaining wall from the first vehicle, are denoted as matrix [M, P, AB] (comprehensive data matrix). By calculating the input matrix [M, P, AB] using a 3D artificial intelligence model, the high-precision curve AB of the retaining wall can be obtained. The high-precision curve data AB of the retaining wall can then be updated and labeled on the spoil heap unloading area map.

[0037] By dividing the open-pit mine map into static maps, road driving area maps, loading area maps, spoil heap unloading area maps, and obstacle maps, it is easier to adopt corresponding map construction and update methods for different areas, which is beneficial for building high-precision maps. Furthermore, the spoil heap unloading area focuses on determining the precise location of retaining walls. Sensor data collected by unmanned trucks during their movement in the spoil heap unloading area is used to construct a coarse curve for the retaining wall, forming a data matrix corresponding to the unmanned truck. Multiple data matrices are then used to construct a comprehensive data matrix, which is further processed using a 3D artificial intelligence model to generate the final retaining wall curve. This method results in a high-precision retaining wall curve, which in turn facilitates the construction of a high-precision open-pit mine map. This allows various mining trucks to operate according to the map, improving operational efficiency and preventing safety accidents.

[0038] Furthermore, by using the unmanned trucks currently in operation as a mobile sensor network to continuously and collaboratively perceive the environment, not only is the real-time performance and efficiency of map updates for the spoil heap unloading area improved, ensuring that the unmanned trucks always make decisions based on the latest environmental information, but more importantly, it enables the newly opened spoil heap unloading area to quickly build a usable high-precision map "from scratch".

[0039] In some examples, after any of the maps—road driving area map, loading area map, spoil heap unloading area map, and obstacle map—is updated, an update notification is sent to vehicles operating in the open-pit mine so that the vehicles can download the latest version of any map indicated in the update notification. The vehicles are configured to initiate a self-check, which involves: upon vehicle startup, requesting the latest version number of each of the road driving area map, loading area map, spoil heap unloading area map, and obstacle map from the cloud; determining the local map that needs updating based on the latest version number; and downloading the latest version of the local map from the cloud. The vehicles are also configured to periodically perform self-checks according to a preset check duration.

[0040] In the complex communication environment of mining areas, network outages and frequent vehicle starts and stops are common occurrences. A multi-layered, complementary, and reliable update strategy is formed by combining three mechanisms: "proactive push," "initiation self-check," and "periodic checks." Proactive push ensures that map updates are distributed to vehicles with minimal latency when network conditions are good. Initiation self-check ensures that vehicles are synchronized with the latest map before each trip, avoiding the risk of carrying outdated maps. Periodic checks serve as a backup for the first two mechanisms, compensating for missed push messages or maintaining synchronization during long-term operation. This design fully considers the actual challenges of industrial sites, ensuring that multiple vehicles use the latest and consistent map versions under complex conditions. This is the foundation for achieving multi-vehicle collaborative scheduling and avoiding planning conflicts caused by map version differences (such as two vehicles being planned to the same location based on different maps).

[0041] In some embodiments, the updating process of the spoil heap unloading area map can be achieved through methods such as... Figure 2 The steps shown are implemented as follows: S201-S204.

[0042] S201 receives sensor data in real time from unmanned trucks operating in the spoil heap unloading area. The sensor data includes first pose data and first laser point cloud data.

[0043] Autonomous trucks operating in the spoil heap unloading area utilize onboard sensors to acquire and upload sensor data to the cloud in real time. These sensors may include 3D LiDAR and positioning and attitude detection devices. In one example, the sensors may include 3D LiDAR, positioning and attitude detection devices, and multiple imaging devices. The images captured by the imaging devices are used for visual perception and auxiliary verification.

[0044] After the driverless truck starts operating in the spoil heap unloading area, its sensors begin to periodically detect the surrounding environment and generate sensor data. The sensor data is then packaged and uploaded to the cloud (if the network is unstable, it is first stored in a local cache queue and retransmitted after the network is restored). The cloud receives the sensor data sent by the driverless truck operating in the spoil heap unloading area in real time.

[0045] S202, a three-dimensional point cloud model of the spoil heap unloading area is constructed based on sensor data.

[0046] In some examples, constructing a 3D point cloud model of the spoil heap unloading area based on sensor data can include fusing sensor data sent by various unmanned trucks operating in the spoil heap unloading area to obtain a dynamically updated 3D point cloud model of the spoil heap unloading area.

[0047] The method of fusing sensor data sent by various unmanned trucks operating in the spoil heap unloading area allows for the fusion of the first laser point cloud data based on the first pose data of each unmanned truck, resulting in the 3D point cloud model. Since the unmanned trucks operating in the spoil heap unloading area send sensor data to the cloud in real time, this 3D point cloud model is essentially dynamically updated.

[0048] In some examples, the coordinate system of the first laser point cloud data is first transformed using the first pose data to the global map coordinate system. Then, all point cloud data transformed to the global map coordinate system over a period of time are fused together to obtain the 3D point cloud model.

[0049] During the fusion process, ICP (Iterative Closest Point) or NDT (Normal Distributions Transform) algorithms can be used to finely align adjacent point cloud frames and eliminate drift caused by minor positioning errors.

[0050] In some cases, the directly fused 3D point cloud model is a preliminary 3D point cloud model. Further denoising and downsampling of this model yields the final 3D point cloud model. This process can involve using a statistical outlier removal filter to eliminate noise and voxel grid downsampling to reduce point cloud density and improve processing efficiency.

[0051] S203 extracts the unloading platform edge, leveled area, and unloading point from the 3D point cloud model in the unloading area of ​​the spoil heap.

[0052] In some cases, the edges of the unloading platform in the unloading area of ​​the spoil heap can be extracted through point cloud elevation change detection.

[0053] The leveled area includes areas that have been leveled by bulldozers and are safe for trucks to drive and turn around. In some examples, the methods for extracting the leveled area from the 3D point cloud model may include: extracting the leveled area from the 3D point cloud model using geometric feature analysis methods; or extracting the leveled area from the 3D point cloud model using deep learning semantic segmentation methods.

[0054] For example, the geometric feature analysis method is based on normal vector and elevation variance analysis. The principle is that the area leveled by a bulldozer is usually a relatively flat plane. The normal vector (a vector perpendicular to the surface) of each point in the point cloud and the elevation variance of its local neighborhood points can be calculated. An elevation variance threshold is set, and areas with elevation variances below this threshold are extracted to obtain the "leveled area." Furthermore, the direction of the normal vector can be used to further filter out non-horizontal planes such as steep slopes.

[0055] Calculate the normal vector: Estimate the normal vector of the point cloud. In a leveled area, the normal vectors of points will be highly consistent (approximately vertically upward). Calculate the elevation variance: For each point, calculate the variance between the elevation values ​​of all points within a certain radius and the average elevation. The smaller the variance, the flatter the area.

[0056] For example, a deep learning model (such as PointNet++, RandLA-Net, KPConv, etc.) that has been trained with a large amount of point cloud data from mining areas can be used to classify the input point cloud point by point. The model outputs a category label for each point, such as: ground, leveled area, unleveled area, retaining wall, vehicle, etc. Then, all points marked as "leveled area" are directly extracted to form a point cloud cluster, and the area corresponding to the point cloud cluster is determined as the leveled area.

[0057] In some cases, the unloading point is an unloading position determined from the leveled area based on the edge of the unloading platform. It should be noted that the unloading point is not a single point, but rather a position used to indicate where the unloading operation will take place.

[0058] S204. Update the unloading area map of the spoil heap based on the edge of the unloading platform, the leveled area and the unloading point in the spoil heap unloading area. The unloading area map of the spoil heap unloading area also marks the status of each unloading point, which can be one of the following: idle, occupied and reserved.

[0059] By constructing a 3D point cloud model and extracting the edges of the unloading platform, leveled areas, and unloading points, the complex and irregular terrain of the spoil heap is transformed into structured data that computers can understand and process. Labeling each unloading point with its "idle," "occupied," or "reserved" status allows for clear visibility of which unloading points are in operation, available, or planned when dispatching vehicles. This enables precise and efficient task allocation and route planning for multiple trucks, preventing congestion or waiting caused by vehicles converging on the same area and preventing vehicles from accidentally entering dangerous areas.

[0060] In some embodiments, the update process of the loading area map can be achieved through methods such as... Figure 3 The steps shown are implemented as follows: S301-S304.

[0061] S301 receives sensor data from electric shovels and / or bulldozers operating in the loading area in real time. The sensor data includes second pose data and second laser point cloud data.

[0062] Taking the real-time reception of sensor data from electric loaders and bulldozers operating in the loading area as an example, the sensors configured on these vehicles may include: 3D LiDAR and positioning and attitude detection devices. In one example, the sensors may include: 3D LiDAR, positioning and attitude detection devices, and multiple imaging devices.

[0063] For example, a positioning and attitude integrated navigation system can specifically be a high-precision GNSS (Global Navigation Satellite System) / RTK (Real-Time Kinematic) + IMU (Inertial Measurement Unit) integrated navigation system to provide high-precision positioning and attitude.

[0064] The data collection process for electric loaders and bulldozers can be as follows: 1. Initialization: Start all sensor drivers and ensure time synchronization (using PTP or NTP protocol). 2. Create a data frame: Start a loop, for example, executing once every 100 milliseconds; 3. Read sensor data, such as obtaining the precise pose at the current moment, obtaining a frame of scan data from the 3D LiDAR, and obtaining images from each camera; 4. Data Packaging: Encapsulate pose data, a frame of laser point cloud data, multiple images, and the current timestamp into a data packet (e.g., using Protobuf or ROS Message format). 5. Data upload: Data packets are sent to the cloud in real time via the network. If the network is unstable, the data is first stored in a local cache queue and retransmitted after the network is restored.

[0065] The cloud receives real-time sensor data from electric loaders and / or bulldozers operating in the loading area.

[0066] S302, constructs a 3D point cloud model of the loading area based on sensor data.

[0067] In some examples, constructing a 3D point cloud model of the loading area based on sensor data may include fusing sensor data sent from electric loaders and / or bulldozers operating in the loading area to obtain a dynamically updated 3D point cloud model of the loading area.

[0068] The method of fusing sensor data sent by various unmanned trucks operating in the loading area allows for the fusion of second laser point cloud data based on the second pose data of each unmanned truck, resulting in the 3D point cloud model. Since the electric loader and / or bulldozer operating in the loading area send sensor data to the cloud in real time, this 3D point cloud model is essentially dynamically updated.

[0069] This fusion process may include: 1. Data unpacking and alignment: Receive data packets from the message queue and accurately align data from different sensors based on timestamps; 2. Coordinate system transformation: Using the electric shovel pose corresponding to each frame of point cloud, the coordinate system of the second laser point cloud data is transformed to the global map coordinate system; 3. Point cloud stitching and denoising: Accumulate all second laser point cloud data converted to the global map coordinate system over a continuous period (e.g., the past 5 minutes). Use ICP or NDT algorithms to finely align adjacent frames, eliminating drift caused by small GNSS / IMU errors, and forming a high-density, high-precision preliminary 3D point cloud model. Apply a statistical outlier removal filter to remove noise from the preliminary 3D point cloud model, and apply voxel downsampling to reduce the density of the preliminary 3D point cloud model to obtain the 3D point cloud model.

[0070] S303, the three-dimensional point cloud model of the loading area is segmented into ground point cloud model, working face point cloud model, ore pile point cloud model, equipment point cloud model and miscellaneous point cloud model, and the boundary line of the working face is extracted, and the drivable area in the ground corresponding to the ground point cloud model is determined.

[0071] For example, a deep learning-based point cloud segmentation model (such as RandLA-Net) or a traditional geometric algorithm can be used to segment the 3D point cloud model of the loading area into ground point cloud model, working face point cloud model, ore pile point cloud model, equipment point cloud model, and miscellaneous point cloud model.

[0072] In some cases, edge detection algorithms (such as Alpha Shape) are applied to the point cloud model of the working face to extract the precise boundary lines of the loading working face.

[0073] In some cases, planar fitting (such as RANSAC) is performed on the ground point cloud model to determine the drivable area.

[0074] S304. Update the loading area map based on the ore pile point cloud model, equipment point cloud model, miscellaneous point cloud model, working face boundary line, and drivable area.

[0075] By segmenting the 3D point cloud model into different categories such as ground, working face, ore pile, and equipment, it is easy to accurately distinguish where the vehicle can travel (drivable area) and where the loading points are (ore pile and working face). The extracted working face boundary lines and the determined drivable areas can facilitate the planning of safe and efficient approach, parking, and departure paths for unmanned trucks, and help solve the problem of collision avoidance when multiple large pieces of equipment interact closely in narrow and dynamic loading areas.

[0076] In some embodiments, the update process of the road driving area map can be achieved through methods such as... Figure 4 The steps shown are implemented as follows: S401-S405.

[0077] The S401 receives point cloud maps and vehicle trajectories in real time from data collection vehicles operating in the road driving area.

[0078] The data collection vehicle is equipped with a higher-precision IMU and multiple 3D LiDARs at different angles. As the vehicle travels along the road within the designated driving area, its onboard computing unit runs laser SLAM algorithms (such as LIO-SAM) to generate high-precision point cloud maps and vehicle trajectories in real time, which are then transmitted to the cloud.

[0079] S402, based on point cloud intensity and shape recognition, determines the lane markings on the road corresponding to the vehicle trajectory in the point cloud map.

[0080] S403: Extract the boundary lines of the road corresponding to the vehicle trajectory, and determine the road centerline based on the boundary lines.

[0081] In some cases, the boundary lines of the road corresponding to the vehicle trajectory can be extracted based on any road edge detection algorithm.

[0082] In some cases, the centerline of a road can be calculated directly from its boundary lines.

[0083] S404 updates the road topology network of the road travel area based on the road corresponding to the vehicle trajectory.

[0084] Based on the roads corresponding to the vehicle trajectories, road segments and intersections are identified, and then the road segments and intersections are connected to construct a complete road topology network.

[0085] S405 updates the road driving area map based on the updated road topology network, lane markings, boundary lines, and road center lines.

[0086] Mining area roads experience frequent changes in condition (such as lane markings and shoulders) due to heavy vehicle traffic and rain erosion. By continuously collecting data via data collection vehicles and automatically identifying lane markings, extracting road boundaries and centerlines, and updating the road topology network, the accuracy of road information relied upon for navigation and route planning is ensured. The updated road topology network clearly defines road connections, driving directions, and rules (such as intersection priorities), facilitating more scientific and rational vehicle scheduling and route planning. Simultaneously, precise lane marking and road boundary information provides clear tracking references for autonomous trucks, enhancing their stability and safety in complex road conditions. Continuous maintenance of the road network on the map is a fundamental prerequisite for ensuring the efficient and orderly operation of the entire mining logistics system, preventing route planning errors, traffic congestion, and even safety accidents caused by outdated maps.

[0087] In some embodiments, the obstacle map base update process can be performed via, for example... Figure 5 The steps shown are implemented as follows: S501-S503.

[0088] The S501 receives obstacle information in real time from vehicles and / or drones traveling in the open-pit mine corresponding to the open-pit mine map. The obstacle information includes the type, location, size, and sensor data of the obstacle.

[0089] It should be noted that the obstacle map is a dynamic layer that is overlaid on other maps.

[0090] Trucks can detect obstacles not on a map (such as large rocks, puddles, or temporarily parked vehicles) using lidar or visual detection while driving, or newly added static obstacles can be detected by drones during regular aerial photography by detecting image changes. The vehicle or drone then reports this obstacle information to the cloud.

[0091] S502 displays obstacle information to the inspector and receives the inspection results from the inspector.

[0092] The cloud displays obstacle information to the inspector, for example, by highlighting the obstacle to be confirmed on the monitoring interface and attaching the sensor data that detected the obstacle.

[0093] The verifier clicks "Confirm" or "Ignore" to send the corresponding verification result to the cloud. If the verifier clicks "Confirm," the verification result is "Obstacle confirmed"; if the verifier clicks "Ignore," the verification result is "Obstacle not found."

[0094] S503: When the verification result confirms an obstacle, update the obstacle map according to the type, location, and size of the obstacle.

[0095] Temporary obstacles in the mining environment (such as falling rocks, accumulated water, and temporarily parked vehicles) pose a major threat to driving safety. By implementing a process of "vehicle perception and reporting - manual verification and confirmation - map update and distribution," each moving truck becomes a mobile "sentinel," forming a dynamic safety monitoring network covering the entire mining area. The introduction of manual verification effectively utilizes human advantages in image and contextual understanding, filtering out false alarms from sensors (such as shadows) and updating only real and persistent threats to the map. This avoids frequent false alarms or unnecessary detours caused by data noise, ensuring the authority and reliability of obstacle information.

[0096] Example 2: Another embodiment of this application relates to an open-pit mine map construction and updating device. The implementation details of this embodiment's open-pit mine map construction and updating device are described below. The following implementation details are provided for ease of understanding and are not essential for implementing this solution. A schematic diagram of this embodiment's open-pit mine map construction and updating device can be seen as follows: Figure 6 As shown, it includes: a construction module 61, used to construct an open-pit mine map containing a static map, a road driving area map, a loading area map, a spoil heap unloading area map, and an obstacle map; and an update module 62, used to update the road driving area map, loading area map, spoil heap unloading area map, and obstacle map in real time. The spoil heap unloading area map includes retaining wall curves, which are obtained by processing a comprehensive data matrix using a 3D artificial intelligence model. The comprehensive data matrix is ​​obtained by merging multiple data matrices, each of which includes multiple coarse curves of retaining walls. Each coarse curve of a retaining wall is determined based on a frame of sensor data collected during the movement of the unmanned truck. The movement process is the process from the unmanned truck entering the spoil heap unloading area to moving to the corresponding collision threshold. The sensor data includes image data and laser point cloud data. One data matrix corresponds to one unmanned truck.

[0097] In some optional embodiments, the update module 62 is used to receive sensor data sent in real time from the unmanned truck operating in the spoil heap unloading area. The sensor data includes first pose data and first laser point cloud data. Based on the sensor data, a three-dimensional point cloud model of the spoil heap unloading area is constructed. The unloading platform edges, leveled areas, and unloading points in the spoil heap unloading area are extracted from the three-dimensional point cloud model. Based on the unloading platform edges, leveled areas, and unloading points in the spoil heap unloading area, the spoil heap unloading area map is updated. The spoil heap unloading area map also marks the status of each unloading point, which is one of idle, occupied, or reserved.

[0098] In some optional embodiments, the update module 62 is used to receive sensor data from electric loaders and / or bulldozers operating in the loading area in real time, the sensor data including second pose data and second laser point cloud data; construct a three-dimensional point cloud model of the loading area based on the sensor data; segment the three-dimensional point cloud model of the loading area into a ground point cloud model, a working face point cloud model, a ore pile point cloud model, an equipment point cloud model, and a miscellaneous point cloud model, and extract the boundary line of the working face, and determine the drivable area in the ground corresponding to the ground point cloud model; update the loading area map according to the boundary line of the ore pile point cloud model, the equipment point cloud model, the miscellaneous point cloud model, the working face, and the drivable area.

[0099] In some optional embodiments, the update module 62 is used to receive point cloud maps and vehicle trajectories sent by the data collection vehicle operating in the road driving area in real time; determine the lane line markings on the road corresponding to the vehicle trajectory in the point cloud map based on point cloud intensity and shape recognition; extract the boundary lines of the road corresponding to the vehicle trajectory, and determine the road centerline based on the boundary lines; update the road topology network of the road driving area according to the road corresponding to the vehicle trajectory; and update the road driving area map according to the updated road topology network, lane line markings, boundary lines and road centerlines.

[0100] In some optional embodiments, the update module 62 is used to receive obstacle information sent by vehicles and / or drones traveling in the open-pit mine corresponding to the open-pit mine map in real time. The obstacle information includes the type, location, size and sensor data of the obstacle; display the obstacle information to the verifier and receive the verification result fed back by the verifier; when the verification result confirms the obstacle, update the obstacle map according to the type, location and size of the obstacle.

[0101] In some optional embodiments, the device further includes a synchronization module 63 for sending an update notification to vehicles operating in the open-pit mine after any of the maps—the road driving area map, the loading area map, the spoil heap unloading area map, and the obstacle map—has been updated, so that the vehicles can download the latest version of any map indicated in the update notification.

[0102] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed in this application; however, this does not mean that other units are absent in this embodiment.

[0103] Example 3: Another embodiment of this application relates to an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the open-pit mine map construction and updating method in the above embodiments.

[0104] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0105] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0106] Example 4: Another embodiment of this application relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0107] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0108] Example 5: Another embodiment of this application relates to a computer program product, including a computer program that, when executed by a processor, implements the above-described method embodiments.

[0109] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A method for constructing and updating open-pit mine maps, characterized in that, include: Construct an open-pit mine map that includes a static map, a road driving area map, a loading area map, a spoil dump unloading area map, and an obstacle map, and update the road driving area map, loading area map, spoil dump unloading area map, and obstacle map in real time; The unloading area map of the spoil heap includes retaining wall curves, which are obtained by processing a comprehensive data matrix using a 3D artificial intelligence model. The comprehensive data matrix is ​​obtained by merging multiple data matrices, each of which includes coarse curves of multiple retaining walls. The coarse curve of each retaining wall is determined based on a frame of sensor data collected during the movement of the unmanned truck. The movement process is the process from the unmanned truck entering the spoil heap unloading area to moving to the corresponding collision critical point. The sensor data includes image data and laser point cloud data. One data matrix corresponds to one unmanned truck.

2. The method for constructing and updating open-pit mine maps according to claim 1, characterized in that, The update process for the spoil heap unloading area map includes: The system receives sensor data in real time from unmanned trucks operating in the unloading area of ​​the spoil heap. The sensor data includes first pose data and first laser point cloud data. A three-dimensional point cloud model of the spoil heap unloading area is constructed based on the sensor data. Extract the unloading platform edge, leveled area and unloading point in the unloading area of ​​the spoil heap from the three-dimensional point cloud model; Based on the unloading platform edges, leveled areas, and unloading points in the spoil heap unloading area, the spoil heap unloading area map is updated. The spoil heap unloading area map also marks the status of each unloading point, which can be one of the following: idle, occupied, or reserved.

3. The method for constructing and updating open-pit mine maps according to claim 1, characterized in that, The update process of the loading area map includes: The system receives sensor data from electric loaders and / or bulldozers operating in the loading area in real time, including second pose data and second laser point cloud data. A three-dimensional point cloud model of the loading area is constructed based on the sensor data; The three-dimensional point cloud model of the loading area is segmented into a ground point cloud model, a working face point cloud model, a ore pile point cloud model, an equipment point cloud model, and a miscellaneous point cloud model. The boundary line of the working face is extracted, and the drivable area in the ground corresponding to the ground point cloud model is determined. Update the loading area map based on the ore pile point cloud model, the equipment point cloud model, the miscellaneous point cloud model, the boundary line of the working face, and the drivable area.

4. The method for constructing and updating open-pit mine maps according to claim 1, characterized in that, The update process for the road driving area map includes: It receives point cloud maps and vehicle trajectories in real time from data collection vehicles operating in the road driving area; Based on point cloud intensity and shape recognition, lane line markings on the road corresponding to the vehicle trajectory in the point cloud map are determined; Extract the boundary lines of the roads corresponding to the vehicle trajectories, and determine the road centerline based on the boundary lines; Update the road topology network of the road travel area based on the road corresponding to the vehicle trajectory; The road driving area map is updated based on the updated road topology network, the lane markings, the boundary lines, and the road centerline.

5. The method for constructing and updating open-pit mine maps according to claim 1, characterized in that, The update process of the obstacle map base includes: The system receives obstacle information in real time from vehicles and / or drones traveling in the open-pit mine corresponding to the open-pit mine map. The obstacle information includes the type, location, size, and sensor data of the obstacles. The obstacle information is displayed to the inspector, and the inspection results are received from the inspector. When the verification result confirms an obstacle, the obstacle map is updated according to the type, location, and size of the obstacle.

6. The method for constructing and updating open-pit mine maps according to any one of claims 1-5, characterized in that, Also includes: Once any one of the road driving area map, loading area map, spoil heap unloading area map, and obstacle map has been updated, an update notification is sent to the vehicles operating in the open-pit mine so that the vehicles can download the latest version of any one of the maps indicated in the update notification. The vehicle is configured to perform a startup self-check, which involves: responding to the vehicle's startup by requesting the latest version numbers of the road driving area map, loading area map, spoil heap unloading area map, and obstacle map from the cloud, determining the local map that needs to be updated based on the latest version numbers, and downloading the latest version of the local map from the cloud. The vehicle is also configured to perform the startup self-test periodically according to a preset check duration.

7. An open-pit mine map construction and updating device, characterized in that, include: The building module is used to build an open-pit mine map that includes a static map, a road driving area map, a loading area map, a spoil heap unloading area map, and an obstacle map; The update module is used to update the road driving area map, loading area map, spoil heap unloading area map and obstacle map in real time; The unloading area map of the spoil heap includes retaining wall curves, which are obtained by processing a comprehensive data matrix using a 3D artificial intelligence model. The comprehensive data matrix is ​​obtained by merging multiple data matrices, each of which includes coarse curves of multiple retaining walls. The coarse curve of each retaining wall is determined based on a frame of sensor data collected during the movement of the unmanned truck. The movement process is the process from the unmanned truck entering the spoil heap unloading area to moving to the corresponding collision critical point. The sensor data includes image data and laser point cloud data. One data matrix corresponds to one unmanned truck.

8. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the open-pit mine map construction and updating method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the open-pit mine map construction and updating method as described in any one of claims 1 to 6.

10. A computer program product comprising a computer program that, when executed by a processor, implements the open-pit mine map construction and updating method according to any one of claims 1 to 6.

Images