Mine obstacle detection method, device and equipment

Abstract

This application discloses a method, apparatus, and equipment for obstacle detection in mining areas, relating to the field of autonomous driving technology. The method includes: acquiring first point cloud data; performing ground segmentation on the first point cloud data to obtain non-ground point cloud data; performing height filtering on the non-ground point cloud data to obtain second point cloud data; ensuring that the first height value corresponding to each first point cloud in the second point cloud data is within a preset height range; clustering the second point cloud data to obtain multiple first clusters; performing feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, thus obtaining multiple target obstacle information; the target obstacle information includes at least one of the target obstacle's center position, target size, and target distance; the target distance is used to characterize the distance between the target obstacle and the nearest point of the target vehicle. According to the embodiments of this application, this method can improve the accuracy of obstacle detection.

Description

Technical Field

[0001] This application belongs to the field of autonomous driving technology, specifically relating to a method, device, and equipment for detecting obstacles in mining areas. Background Technology

[0002] With the rapid development of unmanned mining truck (UAV) technology, point cloud target perception technology based on 3D LiDAR has become the core of the environmental perception field. In real and complex driving environments (such as mining areas and unstructured roads), accurately detecting surrounding obstacles (such as rocks, mounds, equipment, etc.) is a prerequisite for ensuring the safety of system decisions.

[0003] In related technologies, obstacles in mining areas are typically detected using the following method: first, the collected lidar point cloud data is roughly segmented to remove points on flat ground; then, a clustering algorithm (such as a region growing algorithm based on Euclidean distance or normal angle) is initiated in the remaining non-ground point cloud; and finally, obstacle information is determined based on the formed clusters. However, the above method suffers from low accuracy in obstacle detection. Summary of the Invention

[0004] The technical problem to be solved by this application is to provide a method, apparatus and equipment for detecting obstacles in mining areas, which addresses the above-mentioned shortcomings of the existing technology. Using this method, the accuracy of obstacle detection can be improved.

[0005] In a first aspect, embodiments of this application provide a method for detecting obstacles in a mining area, including: Obtain the first point cloud data; The first point cloud data is segmented into ground data to obtain non-ground point cloud data. The non-ground point cloud data is height filtered to obtain the second point cloud data; the first height value corresponding to each first point cloud in the second point cloud data is within the preset height range. Clustering the second point cloud data yields multiple first clusters; Feature analysis is performed on each first cluster to obtain target obstacle information corresponding to each first cluster, thus obtaining multiple target obstacle information; the target obstacle information includes at least one of the target center position, target size and target distance of the target obstacle; the target distance is used to characterize the distance between the target obstacle and the nearest point of the target vehicle.

[0006] In some implementations of the first aspect, acquiring the first point cloud data includes: Acquire lidar point cloud data; The lidar point cloud data is preprocessed to obtain the first point cloud data; the preprocessing includes at least one of format conversion processing and voxel filtering downsampling processing.

[0007] In some embodiments of the first aspect, ground segmentation is performed on the first point cloud data to obtain non-ground point cloud data, including: Obtain the second altitude value corresponding to each second point cloud in the first point cloud data; Multiple third point clouds are selected from all second point clouds; the third height value corresponding to the third point cloud is greater than or equal to the sum of a preset distance threshold and the minimum second height value; All third point clouds were identified as non-ground point cloud data.

[0008] In some embodiments of the first aspect, ground segmentation is performed on the first point cloud data to obtain non-ground point cloud data, including: A random sampling consensus algorithm is used to perform ground segmentation on the first point cloud data to obtain non-ground point cloud data.

[0009] In some implementations of the first aspect, the second point cloud data is clustered to obtain multiple first clusters, including: A density-based spatial clustering algorithm with noise is used to cluster the second point cloud data, resulting in multiple first clusters.

[0010] In some embodiments of the first aspect, before performing feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, and before obtaining multiple target obstacle information, the method further includes: Based on a preset range of points, the first cluster is filtered to obtain a second cluster; the number of points in the second cluster is within the preset range. Feature analysis is performed on each first cluster to obtain target obstacle information corresponding to each first cluster, thus obtaining multiple target obstacle information, including: Feature analysis is performed on each second cluster to obtain target obstacle information corresponding to each second cluster, thereby obtaining multiple target obstacle information.

[0011] In some embodiments of the first aspect, after performing feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, and obtaining multiple target obstacle information, the method further includes: For each target obstacle, the information is converted into target visualization tags to obtain multiple target visualization tags; Combine all target visualization tags into a target tag array message and publish the target tag array message for visualization tools to visualize.

[0012] In some embodiments of the first aspect, after performing feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, and obtaining multiple target obstacle information, the method further includes: The minimum target distance is encapsulated into a target rock distance message, and the rock distance message is published through the target topic for the upper-level decision control module to subscribe to and use.

[0013] Based on the same inventive concept, in a second aspect, embodiments of this application also provide a mining area obstacle detection device, comprising: The first acquisition module is used to acquire the first point cloud data; The first segmentation module is used to perform ground segmentation on the first point cloud data to obtain non-ground point cloud data. The first filtering module is used to perform height filtering on the non-ground point cloud data to obtain the second point cloud data; the first height value corresponding to each first point cloud in the second point cloud data is within a preset height range. The first clustering module is used to cluster the second point cloud data to obtain multiple first clusters; The first analysis module is used to perform feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, so as to obtain multiple target obstacle information; the target obstacle information includes at least one of the target center position, target size and target distance of the target obstacle; the target distance is used to characterize the distance between the target obstacle and the nearest point of the target vehicle.

[0014] Based on the same inventive concept, in a third aspect, embodiments of this application provide an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores one or more computer programs executable by the at least one processor, the one or more computer programs being executed by the at least one processor to enable the at least one processor to perform the above-described method for detecting obstacles in mining areas.

[0015] Based on the same inventive concept, in a fourth aspect, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described method for detecting obstacles in mining areas.

[0016] Based on the same inventive concept, in a fifth aspect, embodiments of this application provide a computer program product that includes computer-readable code or a non-volatile computer-readable storage medium carrying computer-readable code. When the computer-readable code is run in the processor of an electronic device, the processor in the electronic device executes the above-described method for detecting obstacles in mining areas.

[0017] According to the mining area obstacle detection method, apparatus and equipment provided in the embodiments of this application, by performing height filtering on non-ground point cloud data, the first height value corresponding to each first point cloud in the obtained second point cloud data is within a preset height range. This can not only exclude small gravel that does not affect vehicle driving, but also filter out non-stone objects such as vehicles and large equipment, so that the detection focuses on the target obstacles that truly threaten driving safety, thereby improving the accuracy of obstacle detection.

[0018] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the present application and form part of the specification. They are used together with the embodiments of the present application to explain the application and do not constitute a limitation thereof. The above and other features and advantages will become more apparent to those skilled in the art from the detailed example embodiments described with reference to the accompanying drawings, in which: Figure 1 This illustration shows a flowchart of a method for detecting obstacles in a mining area provided in an embodiment of this application. Figure 2 This is a visualization diagram of point cloud data obtained by lidar detection of an unmanned mining truck according to an embodiment of this application; Figure 3 This illustration shows another flowchart of the obstacle detection method in mining areas provided in an embodiment of this application; Figure 4 This illustration shows a structural schematic diagram of a mine obstacle detection device provided in an embodiment of this application; Figure 5 This illustration shows a structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions of this application, exemplary embodiments of this application are described below in conjunction with the accompanying drawings, including various details of the embodiments of this application to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0021] Where there is no conflict, the various embodiments of this application and the features thereof may be combined with each other.

[0022] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0024] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and this application, and will not be interpreted as having an idealized or overly formal meaning, unless expressly so defined herein.

[0025] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0026] The obstacle detection method for mining areas provided in this application is applicable to unmanned mining trucks traveling on mining roads. This method can be executed by an obstacle detection device, electronic equipment, or a mining obstacle detection system. The following description uses the execution of this method by a mining obstacle detection system as an example.

[0027] For example, the obstacle detection system in the mining area may include hardware modules and software modules, and through the collaboration of multiple modules, point cloud acquisition, ground segmentation, target obstacle identification and danger distance assessment can be achieved.

[0028] Specifically, the hardware module may include a lidar unit, a communication and computing unit, and a visualization unit. The lidar unit employs a semi-solid-state lidar, rigidly fixed to the center of the front bumper of the unmanned mining vehicle via a custom stainless steel bracket. The vertical height of the radar center from the ground is precisely controlled at 2.0m (error ≤ ±5cm), with a 3° downward tilt angle to ensure beam coverage of the road area within 0.5-50m in front of the vehicle. Polyurethane damping pads (20mm diameter, 5mm thickness, Shore A hardness 60) are added to the connection point between the bracket and the vehicle body to reduce vibration transmission when the vehicle is traveling on unpaved roads. The lidar is used to collect point cloud data of the road in front of the vehicle, outputting point cloud messages containing x, y, and z coordinate information (i.e., lidar point cloud data). Its core function is to provide raw point cloud data for the stone detection system, used for subsequent ground segmentation, height filtering, and clustering identification.

[0029] Communication and Computing Unit: The onboard computing unit runs Ubuntu 20.04+ Robot Operating System (ROS) Noetic. It communicates with the LiDAR driver node via ROS topics and subscribes to point cloud topics to obtain point cloud data in real time. The computing unit is responsible for executing the stone detection algorithm and publishing information such as stone position, size, and distance.

[0030] Visualization Unit: The 3D visualization tool (RViz) displays the stone detection results in real time, aiding in debugging and monitoring. In RViz's "base_link" coordinate system, each detected stone is marked with a red semi-transparent cube. The cube's position corresponds to the center point of the stone, and its size corresponds to the actual size of the stone plus a 0.1m margin. The marker's lifespan is set to 0.5 seconds to ensure that the marker updates in real time with the point cloud, and old markers disappear automatically.

[0031] Specifically, the software module can be used to execute the obstacle detection method in the mining area provided in the embodiments of this application.

[0032] The following describes the obstacle detection method in mining areas provided by the embodiments of this application.

[0033] like Figure 1 As shown in the embodiment of this application, the obstacle detection method in the mining area includes steps S110 to S150.

[0034] S110, Obtain the first point cloud data.

[0035] S120. Perform ground segmentation on the first point cloud data to obtain non-ground point cloud data.

[0036] S130. Perform height filtering on the non-ground point cloud data to obtain the second point cloud data; the first height value corresponding to each first point cloud in the second point cloud data is within the preset height range.

[0037] S140. Cluster the second point cloud data to obtain multiple first clusters.

[0038] S150. Perform feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, so as to obtain multiple target obstacle information; the target obstacle information includes at least one of the target center position, target size and target distance of the target obstacle; the target distance is used to characterize the distance between the target obstacle and the nearest point of the target vehicle.

[0039] According to the obstacle detection method in the mining area provided in the embodiments of this application, by performing height filtering on the non-ground point cloud data, the first height value corresponding to each first point cloud in the obtained second point cloud data is within a preset height range. This method can exclude small gravel that does not affect vehicle driving, and filter out non-stone objects such as vehicles and large equipment, so that the detection focuses on the target obstacles that truly threaten driving safety, thereby improving the accuracy of obstacle detection.

[0040] The specific implementation methods for each of the above steps are described below.

[0041] In step S110, the first point cloud data can be lidar point cloud data or point cloud data obtained after preprocessing lidar point cloud data.

[0042] In some implementations, acquiring the first point cloud data includes: Acquire lidar point cloud data; The lidar point cloud data is preprocessed to obtain the first point cloud data; the preprocessing includes at least one of format conversion processing and voxel filtering downsampling processing.

[0043] In this embodiment, by performing format conversion processing on the lidar point cloud data, invalid points containing null (NaN) values ​​can be filtered out; and / or, by performing voxel filtering downsampling processing on the lidar point cloud data, the density of the lidar point cloud data can be reduced, thereby reducing the computational load of subsequent processing while preserving the key geometric features of the target obstacle.

[0044] For example, LiDAR point cloud data can be subscribed to via a ROS topic, with the topic being / front / rslidar_points.

[0045] For example, LiDAR point cloud data can be preprocessed using a callback function.

[0046] For example, the format conversion process includes converting the point cloud (i.e., the first point cloud data) in Point Cloud 2 format into an N-dimensional (Numpy) array with the format (N, 3), where each row contains the x, y, and z coordinates of the point cloud. Invalid points containing null values ​​are filtered out during the conversion process.

[0047] For example, voxel filtering downsampling processing includes: using the Open3D library to perform voxel filtering downsampling on the point cloud (LiDAR point cloud data, or LiDAR point cloud data after format conversion), with the voxel size set to 0.02 meters. Voxel filtering divides the three-dimensional space into fixed-size cubes (voxels), retaining only one centroid point within each voxel, thereby reducing the point cloud density and lowering the computational load of subsequent processing. Simultaneously, voxel filtering smooths noise while preserving the overall geometry of the stone. The downsampled point cloud retains the key features of the stone while significantly reducing the number of points, improving processing efficiency.

[0048] In this embodiment, voxel filtering downsampling (voxel size 0.02 meters) is used to reduce the density of the original point cloud by 60%-80%, which greatly reduces the computational load of subsequent processing while preserving the key geometric features of the stone.

[0049] Understandably, when preprocessing includes format conversion and voxel filtering downsampling, the LiDAR point cloud data can be format converted first to obtain the format-converted LiDAR point cloud data; then, the format-converted LiDAR point cloud data can be voxel filtering downsampling to obtain the first point cloud data.

[0050] In step S120, ground segmentation is a key step in target obstacle detection. Accurate ground segmentation can effectively remove road surface interference, allowing subsequent processing to focus on objects that may be rocks (i.e., target obstacles) that are off the ground.

[0051] Ground segmentation of the first point cloud data yields ground point cloud data and non-ground point cloud data.

[0052] In some implementations, ground segmentation is performed on the first point cloud data to obtain non-ground point cloud data, including: The Random Sampling Consensus (RANSAC) algorithm is used to perform ground segmentation on the first point cloud data to obtain non-ground point cloud data.

[0053] In this embodiment, the RANSAC algorithm is used to perform ground segmentation on the first point cloud data, which can accurately separate ground point cloud data from non-ground point cloud data.

[0054] For example, the RANSAC algorithm is used to segment the ground in the first point cloud data, including: Plane model fitting: three points are randomly selected from the first point cloud data, a plane model is fitted, the distances from other points to the plane are calculated, the number of interior points with a distance less than a first distance threshold (e.g., 0.2 meters) is counted, and the iteration is repeated a preset number of times (e.g., 100 times). The plane with the most interior points is selected as the optimal ground plane; Ground point extraction: points with a distance less than 0.2 meters from the optimal ground plane are marked as ground points, and the remaining points are marked as non-ground points, to obtain ground point cloud data and non-ground point cloud data. Among them, ground points include road surfaces, flat areas, etc., and non-ground points include rocks, vehicles, pedestrians, vegetation, etc.

[0055] In this embodiment, the RANSAC algorithm is optimized by limiting the maximum number of iterations (100 times) and the first distance threshold, thereby controlling the computation time while ensuring segmentation accuracy.

[0056] It should be noted that setting the ground segmentation distance threshold (i.e., the first distance threshold) to 0.2 meters can effectively filter out road surface interference and avoid misjudging raised ground as target obstacles such as rocks.

[0057] Of course, the first distance threshold can also be set according to the actual situation, and is not limited here.

[0058] In some implementations, ground segmentation is performed on the first point cloud data to obtain non-ground point cloud data, including: Obtain the second altitude value corresponding to each second point cloud in the first point cloud data; Multiple third point clouds are selected from all second point clouds; the third height value corresponding to the third point cloud is greater than or equal to the sum of a preset distance threshold and the minimum second height value; All third point clouds were identified as non-ground point cloud data.

[0059] In other words, the minimum Z value (i.e. the minimum second altitude value) of all second point clouds in the first point cloud data is calculated. Points with Z values ​​less than (minimum Z value + preset distance threshold) are taken as ground points to obtain ground point cloud data; points with Z values ​​greater than or equal to (minimum Z value + preset distance threshold) (i.e. third point cloud) are taken as non-ground points to obtain non-ground point cloud data.

[0060] It should be noted that when RANSAC segmentation of the point cloud library (PCL) fails, the simple face segmentation based on height described above can be used as an alternative to ensure that the system can still operate under abnormal conditions.

[0061] In step S130, the segmented non-ground point cloud data is subjected to height filtering to extract non-ground point cloud data with heights within a preset height range.

[0062] For example, the height filtering process includes: height screening, traversing each point cloud in the non-ground point cloud data, extracting its Z coordinate (i.e., the first height value), and determining whether the Z coordinate is within a preset height range (0.4 meters to 1 meter); high point cloud extraction: retaining the first point clouds that meet the height conditions (i.e., the preset height range) to form a high point cloud set (i.e., the second point cloud data). These first point clouds correspond to objects with a height (i.e., the first height value) of 40 centimeters or more but less than 1 meter, which are very likely to be target obstacles such as rocks that pose a threat to the driving of mining trucks. Through the above height filtering, the first point clouds in the height range of 0.4 meters to 1.0 meters are accurately extracted, which not only excludes small gravel (below 0.4 meters) that does not affect driving, but also filters out non-rock objects such as vehicles and large equipment (above 1.0 meters), so that the detection focuses on target obstacles such as rocks that truly threaten driving safety.

[0063] Understandably, height filtering addresses the problem of excessive gravel in mining environments: gravel smaller than 40 centimeters does not cause serious damage to mining cards and can be ignored to reduce computational load; objects larger than 1 meter are mostly vehicles, large equipment, etc., which do not fall within the detection scope of rocks and other obstacles and are also ignored. Through height filtering, the system can focus on the truly important obstacles such as rocks.

[0064] Of course, the preset height range can also be set according to the actual situation, and is not limited here.

[0065] In step S140, the height-filtered second point cloud data is clustered, grouping discrete points belonging to the same target obstacle, such as a rock, into a single cluster. The first cluster is the cluster of discrete points belonging to the same target obstacle.

[0066] In some implementations, the second point cloud data is clustered to obtain multiple first clusters, including: The density-based spatial clustering with noise (DBSCAN) algorithm is used to cluster the second point cloud data, resulting in multiple first clusters.

[0067] In this embodiment, the DBSCAN algorithm is used to aggregate discrete point clouds (i.e., second point cloud data) belonging to the same target obstacle such as a stone into independent clusters (i.e., first clusters), which can accurately segment adjacent target obstacles such as stones and avoid misjudgment due to adhesion.

[0068] For example, the process of clustering the second point cloud data using the DBSCAN algorithm includes: Clustering parameter settings: The neighborhood radius is set to 0.2 meters, meaning that points within 20 centimeters are considered neighbors; the minimum number of samples is set to 10 points, meaning that a cluster (i.e., the first cluster) must include at least 10 points, and clusters with fewer than 10 points are considered noise and ignored. Clustering execution: The DBSCAN algorithm starts from any unvisited point and finds all points within its eps neighborhood (i.e., the first point cloud). If the number of points in the neighborhood is greater than or equal to the minimum cluster size (i.e., the minimum number of samples), these points form a new cluster, and the cluster is recursively expanded; otherwise, the point is marked as noise. The algorithm traverses all points and eventually obtains multiple clusters (i.e., the first clustering clusters), each cluster corresponding to a potential independent stone (i.e., the target obstacle).

[0069] It should be noted that setting the neighborhood radius to 0.2 meters can accurately separate adjacent stones and avoid misjudgment due to adhesion.

[0070] In this embodiment, the DBSCAN clustering algorithm optimizes parameters (eps=0.2 meters, min_samples=10) for the point cloud density characteristics of the mining area, thereby improving the processing speed while ensuring the clustering effect.

[0071] Of course, the neighborhood radius and minimum number of samples can be set according to the actual situation, and are not limited here.

[0072] It should be noted that DBSCAN clustering can automatically identify the first cluster of any shape without needing to pre-specify the number of the first cluster, making it very suitable for detecting irregularly shaped rocks and other obstacles in mining areas.

[0073] In step S150, feature analysis is performed on each first cluster to identify the rocks (i.e. target obstacles) and assess their degree of danger.

[0074] For example, the target obstacle may include rocks, etc.

[0075] For example, the target obstacle feature calculation includes: for each first cluster, calculating the following features: target height range, target bounding box range, and target center point location, etc.

[0076] For example, the target distance is the closest dangerous distance to the target obstacle, which is a key indicator for the decision control module to determine whether avoidance is necessary. Specifically, the Euclidean distance from each point in the first cluster to the lidar sensor (located at the origin of the vehicle coordinate system) is calculated, and the minimum value of this Euclidean distance is taken as the target distance. This target distance represents the distance between the target obstacle and the closest point of the target vehicle.

[0077] Furthermore, it can traverse all detected target obstacles and record the minimum target distance as the global danger distance for this detection, which will be used for subsequent releases.

[0078] In some implementations, before performing feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, and before obtaining multiple target obstacle information, the method further includes: Based on a preset range of points, the first cluster is filtered to obtain a second cluster; the number of points in the second cluster is within the preset range. Feature analysis is performed on each first cluster to obtain target obstacle information corresponding to each first cluster, thus obtaining multiple target obstacle information, including: Feature analysis is performed on each second cluster to obtain target obstacle information corresponding to each second cluster, thereby obtaining multiple target obstacle information.

[0079] In this embodiment, by using a preset range of points, the first cluster with too few filtered points may be a noisy or too small first cluster, and the first cluster with too many filtered points may be a vehicle or large equipment first cluster, so as to further screen out candidate targets that meet the characteristics of the target obstacle.

[0080] It should be noted that the preset point range can be set according to the actual situation and is not limited here. For example, the preset point range is between 10 and 500 points.

[0081] In some implementations, feature analysis is performed on each first cluster to obtain target obstacle information corresponding to each first cluster. After obtaining multiple target obstacle information, the method further includes: For each target obstacle, the information is converted into target visualization tags to obtain multiple target visualization tags; Combine all target visualization tags into a target tag array message and publish the target tag array message for visualization tools to visualize.

[0082] For example, target visualization markers may include 3D bounding box markers. The process of creating 3D bounding box markers includes: creating a marker message (visualization_msgs / Marker) of type cube in a visualization message packet for each target obstacle, representing the 3D bounding box of the target obstacle; and setting individual marker attribute settings.

[0083] For example, the markers of all template obstacles (i.e. target visualization markers) are combined into a target marker array message (MarkerArray), which is published via the topic / rock_markers for display by visualization tools such as the robot operating system RViz.

[0084] In some implementations, feature analysis is performed on each first cluster to obtain target obstacle information corresponding to each first cluster. After obtaining multiple target obstacle information, the method further includes: The minimum target distance is encapsulated into a target rock distance message, and the rock distance message is published through the target topic for the upper-level decision control module to subscribe to and use.

[0085] For example, the minimum target distance detected this time is encapsulated into a custom message called RockDistance, and published through the target topic of rock distance ( / perception / rock_distance) under the perception module for subscription and use by the upper-level decision control module.

[0086] Furthermore, this method can also output logs: the detection results are output in real time on the terminal, including the number of detected target obstacles, the center position of the target, the size of the target, the processing time, etc., which facilitates development and debugging.

[0087] Furthermore, the method also includes: a built-in performance statistics function to monitor processing time in real time. Time consumption statistics: Record the total time consumed from receiving to processing each frame of point cloud, and save the processing time of the most recent 100 frames; Average processing time calculation: Calculate the average processing time of the last 100 frames, and output it to the terminal every 5 seconds, including the processing time of the current frame and the average processing time; Real-time performance guarantee: Through voxel filtering downsampling, optimized algorithm flow, and reasonable parameter settings, the processing time is ensured to be controlled within 100 milliseconds, meeting the real-time requirements of unmanned driving in the mining area.

[0088] Furthermore, before implementing the obstacle detection method in the mining area provided in this application embodiment, parameter configuration and initialization can also be performed.

[0089] Specifically, during the system startup phase, the configuration parameters required for stone detection are obtained from the parameter server, including the ground segmentation distance threshold (i.e., the first distance threshold), the stone height range (i.e., the preset height range), clustering parameters, and voxel filter size. The specific parameter settings are as follows: Ground segmentation distance threshold (ground_distance_threshold): Set to 0.2 meters, meaning points within 20 centimeters of the fitted plane are considered ground. This threshold determines the strictness of ground segmentation; a threshold that is too high will misclassify low-lying rocks as ground, while a threshold that is too low will misclassify undulating ground as non-ground.

[0090] Rock height range: The minimum rock height (rock_height_min) is set to 0.4 meters, which means that only obstacles 40 centimeters or more will be detected, because rocks below 40 centimeters usually will not cause serious damage to the mining truck tires and chassis; the maximum rock height (rock_height_max) is set to 1.0 meter, which ignores objects that are too tall (such as vehicles, pedestrians, and large equipment) to avoid false detections.

[0091] Clustering parameters: The neighborhood radius (cluster_tolerance) is set to 0.2 meters, which represents the maximum distance between two points to determine if they belong to the same object. Points within 20 centimeters are considered to be the same stone. The minimum number of samples (min_cluster_size) is set to 10 points, and clusters with fewer than 10 points are ignored (to filter out noise and objects that are too small). The maximum number of samples (max_cluster_size) is set to 500 points, and clusters with more than 500 points are ignored (to filter out objects that are too large, such as vehicles).

[0092] Voxel filter size (voxel_size): Set to 0.02 meters, which means merging points within a 2-centimeter cube into one point, used for downsampling to reduce computation while preserving the key geometric features of the stone.

[0093] At the same time, the lidar coordinate system is initialized, and the lidar is set to be located at the origin of the vehicle coordinate system for subsequent distance calculation.

[0094] This application embodiment constructs a complete method for detecting stones (i.e. obstacles) on unmanned mining trucks in mining areas. It achieves accurate identification, positioning, and dangerous distance assessment of stones larger than 40 centimeters on the road, effectively solving the problems of low stone detection accuracy and poor real-time performance in complex mining environments, and providing reliable protection for the safe operation of unmanned mining trucks.

[0095] The embodiments of this application significantly improve the accuracy of stone detection through a multi-level and multi-dimensional screening mechanism, effectively solving the problem of missed or false detection of stones in complex mining environments.

[0096] To better understand the obstacle detection method in mining areas provided in the embodiments of this application, the following explanation is provided with specific examples.

[0097] For example, a visualization diagram of the point cloud data obtained by the lidar detection of the unmanned mining truck provided in the embodiments of this application is shown below. Figure 2As shown, the background is black, and the white point cloud and scan lines are clearly presented. The radial concentric rings (representing different distance layers of radar scanning, from near to far) show the point cloud distribution covering the area in front of the vehicle. The bottom center (such as in an unmanned mining vehicle) is the radar mounting point. The overall point cloud data corresponds to the radar's perception range of the road ahead (approximately 0.5-50 meters), which can be used for subsequent ground segmentation, stone detection, and other processing. It intuitively demonstrates the radar's perception effect on the environment ahead and the distribution characteristics of obstacles.

[0098] like Figure 3 As shown in the embodiment of this application, the obstacle detection method in the mining area includes: after starting, firstly receiving the lidar topic / front / islidar / points; then performing voxel filtering downsampling (reducing the amount of computation); subsequently performing RANSAC ground segmentation (within 20cm of the ground is considered the ground), this step is divided into two branches, the left side is the ground point cloud ground_points, and the right side is the non-ground point cloud non_ground_points; then performing height filtering (filtering conditions: 0.4m < height < 1.0m; > 0.4m: excluding small gravel that does not affect driving; < 0.1m: excluding vehicles, large amounts of equipment and other non-stone objects); and then performing DBSCAN clustering (grouping discrete points belonging to the same stone block). The data is clustered into clusters, resulting in cluster 1 (stones) containing N points with height h1, ..., cluster K (stones) containing M points with height h2. Height filtering is then performed (10 <= number of cluster points <= 500). Feature calculation is then performed (for each filtered cluster, the center point and size are calculated). Distance calculation is then performed (the distance from each point in all clusters to the sensor is recorded, and the minimum value is taken as the minimum distance). Visual marker creation is then performed (3D bounding box markers are created for each stone). Finally, the results are published (the `rock_markers` topic (visual markers); ` / perception / rock_detection` (stone distance information); and terminal logs (number of stones, location, height, and time taken)). The process ends, and the system waits for the next frame.

[0099] More specifically, the obstacle detection method in the mining area provided in this application includes: S1: System Deployment S11: Application Scenarios This application describes an unmanned mining truck used in an open-pit coal mine. The vehicle transports ore between the mining area and the spoil heap. The road is an unpaved surface within the mine, frequently littered with loose rocks. When the truck travels at 25-35 km / h, failure to detect and avoid larger rocks (over 40 cm) in time can lead to serious malfunctions such as tire blowouts and chassis damage. A lidar sensor is installed at the front of the vehicle to detect rock obstacles on the road ahead.

[0100] S12: Deployment Plan

[0101] A semi-solid-state lidar is used, fixed to the center front bumper of the vehicle via a stainless steel bracket, 2.0m above the ground, with a 3° downward tilt angle to ensure beam coverage of the road area within 0.5-50m in front of the vehicle. Polyurethane damping pads are added at the connection between the bracket and the vehicle body to reduce vehicle vibration.

[0102] The lidar coordinate system has its origin (O) at the physical center of the lidar: X-axis: Forward is the positive direction (vehicle travel direction) Y-axis: Positive direction is to the left (horizontal offset) Z-axis: Upward is the positive direction (height). S13: Software Environment Configuration The onboard computing unit runs Ubuntu 20.04+ROS Noetic and deploys the stone detection node of this application embodiment. The nodes communicate with each other through ROS topics as shown in Table 1.

[0103] Table 1

[0104] The contents of Table 1 are for illustrative purposes only and are not intended to limit this application.

[0105] S2: Parameter Configuration

[0106] Based on the characteristics of the mining area roads and the type of mining truck, the parameters are configured as shown in Table 2.

[0107] Table 2

[0108] The contents of Table 2 are for illustrative purposes only and are not intended to limit this application.

[0109] S3: Data Processing Flow

[0110] S31: Point cloud reception and preprocessing

[0111] The system receives LiDAR point cloud data at a frequency of 10Hz. First, it converts the ROS point cloud format into a coordinate array, and then performs voxel filtering downsampling. Voxel filtering merges points within a 2cm×2cm×2cm space into a single centroid. The original point cloud has approximately 30,000 points, and after downsampling, it has approximately 6,000-8,000 points, reducing the number of points by more than 70% and significantly reducing the amount of subsequent computation.

[0112] S32: RANSAC Ground Segmentation

[0113] The RANSAC algorithm was used for ground segmentation. Three points were randomly selected to fit a planar model, and the distances from other points to this plane were calculated. Points with a distance less than 0.2m were considered ground points. After 100 iterations, the optimal plane was selected, and the point cloud was divided into ground point cloud and non-ground point cloud. Ground point cloud (road surface) accounted for approximately 60-70%, and non-ground point cloud (stones, vehicles, etc.) accounted for approximately 30-40%.

[0114] S33: Height Filtering

[0115] Height filtering is applied to the non-ground point cloud to extract points with heights between 0.4m and 1.0m. Small pebbles below 0.4m are filtered out (as they do not affect driving), while vehicles, equipment, etc., above 1.0m are filtered out (as they are not considered rocks). The remaining point cloud focuses on potential rock targets.

[0116] S34: DBSCAN Clustering

[0117] DBSCAN clustering was performed on the height-filtered point cloud, with parameters eps=0.2m (neighborhood radius) and min_samples=10 points. The algorithm clusters points within 20cm into a single cluster, ignoring points with a label of -1 as noise. Multiple clusters were obtained after clustering, each corresponding to an independent stone candidate.

[0118] S35: Cluster Filtering and Stone Recognition

[0119] For each cluster, a size filter is applied, retaining clusters with 10-500 points. Clusters with too few points are considered noise and ignored, while clusters with too many points are considered large objects and ignored. Clusters that pass the filter are identified as stones.

[0120] S36: Stone Feature Calculation

[0121] For each identified stone, calculate the following features: Bounding box range: x_min, x_max, y_min, y_max, z_min, z_max Center point locations: center_x, center_y, center_z Stone dimensions: width (x), length (y), height (z) S37: Danger Distance Calculation Calculate the distance from each point in the rock cluster to the lidar sensor (origin of the vehicle coordinate system), and take the minimum value as the nearest danger distance for that rock. Traverse all rocks and record the global minimum distance.

[0122] S4: Visual Markup Creation

[0123] Create 3D bounding box markers for each stone for RViz visualization: Location: Coordinates of the center point of the stone Dimensions: Actual size of the stone plus a 0.1m allowance. Color: Red semi-transparent (r=1.0, g=0.0, b=0.0, a=0.5) Lifecycle: 0.5 seconds, updated in real time with point cloud. S5: Results Release The system publishes the following content per frame: / rock_markers: An array of 3D bounding box markers for all rocks. / perception / rock_distance: Global minimum rock distance and number of rocks Terminal log: Number of stones, minimum distance, processing time (output every 5 seconds) Overall Implementation Effect Through the above examples, the system was implemented in a complex road environment in a mining area: 1. Ground segmentation: The RANSAC ground segmentation algorithm is used with a distance threshold of 0.2 meters to accurately separate ground and non-ground point clouds on the uneven unpaved road surface in the mining area.

[0124] 2. Stone detection: Accurately identifies stones larger than 40cm on the road through height filtering and DBSCAN clustering.

[0125] 3. Distance assessment: Accurately calculate the closest distance from each stone to the vehicle, with an error of ≤0.1 meters.

[0126] 4. Real-time response: Single-frame point cloud processing time ≤ 50 milliseconds, meeting the requirements for 10Hz real-time detection.

[0127] 5. Visual monitoring: The RViz interface displays a red semi-transparent 3D bounding box in real time, clearly marking the position, size and distribution of each stone.

[0128] The embodiments of this application have at least the following beneficial effects: 1. High detection accuracy, effectively identifying stones larger than 40cm on the road. To address the complex situation of varying stone sizes, random distribution, and easy confusion with the ground in mining road environments, this invention employs a detection strategy combining multi-level filtering and clustering to achieve high-precision identification of stones larger than 40cm on roads. The RANSAC ground segmentation algorithm accurately separates ground point clouds from non-ground point clouds, with a ground segmentation distance threshold set at 0.2 meters, effectively filtering out road surface interference and preventing misidentification of protruding ground features as stones. Height filtering precisely extracts point clouds within a height range of 0.4 to 1.0 meters, excluding small gravel (below 0.4 meters) that does not affect driving, and filtering out non-stone objects such as vehicles and large equipment (above 1.0 meters), focusing detection on stones that truly threaten driving safety. The DBSCAN clustering algorithm groups discrete point clouds belonging to the same stone into independent clusters, with a cluster neighborhood radius set at 0.2 meters, accurately segmenting adjacent stones and avoiding misidentification due to overlapping stones. This invention significantly improves the accuracy of stone detection through a multi-level and multi-dimensional screening mechanism, effectively solving the problem of missed or false detection of stones in complex mining environments.

[0129] 2. Excellent real-time performance, meeting the needs of unmanned high-speed driving in mining areas.

[0130] This invention achieves millisecond-level real-time processing of stone detection through optimized algorithm flow and parameter configuration. Voxel filtering downsampling (voxel size 0.02 meters) reduces the original point cloud density by 60%-80%, significantly reducing the computational load of subsequent processing while preserving the key geometric features of the stones. The RANSAC ground segmentation algorithm is optimized by limiting the maximum number of iterations (100 times) and distance thresholds, controlling computation time while ensuring segmentation accuracy. The DBSCAN clustering algorithm optimizes parameters (eps=0.2 meters, min_samples=10) for the point cloud density characteristics of mining areas, improving processing speed while maintaining clustering effectiveness. The system has a built-in performance statistics module that monitors the processing time of each frame of point cloud in real time. The average processing time can be controlled within 50 milliseconds, with a maximum of no more than 100 milliseconds, fully meeting the real-time detection requirements of mining trucks traveling at 30 km / h in mining areas (100 millisecond interval between each frame of point cloud). When a stone is detected, the system can transmit information such as the stone's location, size, and distance to the upper-level decision control module in the next processing cycle, triggering obstacle avoidance or deceleration strategies to effectively avoid collisions caused by detection delays.

[0131] 3. Visual and intuitive, assisting operators in monitoring

[0132] This invention utilizes the RViz visualization tool to create 3D bounding box markers for each detected stone, displaying the stone's position, size, and distribution in real time. Each stone marker is displayed as a red, semi-transparent cube, making it highly visible and easily identifiable. Its size corresponds to the actual size of the stone (plus a 0.1-meter margin), and its position corresponds to the stone's center point. A lifespan of 0.5 seconds ensures the marker updates in real time with the point cloud. Operators can intuitively perceive the distribution of stones on the road ahead through the visualization interface, anticipating potential hazards and providing manual monitoring support in addition to the autonomous driving system's decision-making. Simultaneously, the terminal log outputs in real time the number of detected stones, the position coordinates and height information of each stone, and the processing time, facilitating monitoring of the system's operational status by development and debugging personnel and enabling timely detection of anomalies.

[0133] 4. Parameters are configurable to adapt to different mining area scenarios.

[0134] This invention employs a parametric design, where all key parameters can be dynamically obtained from the ROS parameter server, adapting to the specific needs of different mining areas without code modification. Configurable parameters include: ground segmentation distance threshold, rock height range, cluster neighborhood radius, minimum / maximum cluster size, and voxel filter size. This flexible parameter configuration mechanism enables the invention to quickly adapt to different mining areas, vehicle models, and LiDAR models, significantly improving the system's adaptability and portability.

[0135] The obstacle detection method in the mining area provided in this application can be executed by a mining area obstacle detection system.

[0136] It is understood that the various method embodiments mentioned above in this application can be combined with each other to form combined embodiments without violating the principle and logic. Due to space limitations, this application will not elaborate further. Those skilled in the art will understand that in the above methods of specific implementation, the specific execution order of each step should be determined by its function and possible internal logic.

[0137] Based on the same inventive concept, embodiments of this application also provide a mining area obstacle detection device. For example... Figure 4 As shown, the device includes a first acquisition module 210, a first segmentation module 220, a first filtering module 230, a first clustering module 240, and a first analysis module 250.

[0138] The first acquisition module 210 is used to acquire the first point cloud data.

[0139] The first segmentation module 220 is used to perform ground segmentation on the first point cloud data to obtain non-ground point cloud data.

[0140] The first filtering module 230 is used to perform height filtering on the non-ground point cloud data to obtain the second point cloud data; the first height value corresponding to each first point cloud in the second point cloud data is within a preset height range.

[0141] The first clustering module 240 is used to cluster the second point cloud data to obtain multiple first clusters.

[0142] The first analysis module 250 is used to perform feature analysis on each first cluster to obtain target obstacle information corresponding to each first cluster, so as to obtain multiple target obstacle information; the target obstacle information includes at least one of the target center position, target size and target distance of the target obstacle; the target distance is used to characterize the distance between the target obstacle and the nearest point of the target vehicle.

[0143] According to the mining area obstacle detection device provided in the embodiments of this application, by performing height filtering on non-ground point cloud data, the first height value corresponding to each first point cloud in the obtained second point cloud data is within a preset height range. This can not only exclude small gravel that does not affect vehicle driving, but also filter out non-stone objects such as vehicles and large equipment, so that the detection focuses on the target obstacles that truly threaten driving safety, thereby improving the accuracy of obstacle detection.

[0144] In some implementations, the first acquisition module 210 is specifically used for: Acquire lidar point cloud data; The lidar point cloud data is preprocessed to obtain the first point cloud data; the preprocessing includes at least one of format conversion processing and voxel filtering downsampling processing.

[0145] In some implementations, the first segmentation module 220 is specifically used for: Obtain the second altitude value corresponding to each second point cloud in the first point cloud data; Multiple third point clouds are selected from all second point clouds; the third height value corresponding to the third point cloud is greater than or equal to the sum of a preset distance threshold and the minimum second height value; All third point clouds were identified as non-ground point cloud data.

[0146] In some implementations, the first segmentation module 220 is specifically used for: A random sampling consensus algorithm is used to perform ground segmentation on the first point cloud data to obtain non-ground point cloud data.

[0147] In some implementations, the first clustering module 240 is specifically used for: A density-based spatial clustering algorithm with noise is used to cluster the second point cloud data, resulting in multiple first clusters.

[0148] In some embodiments, the device may further include: The first filtering module is used to filter the first cluster based on a preset range of points to obtain a second cluster; the number of points in the second cluster is within the preset range. The first analysis module 250 is specifically used for: Feature analysis is performed on each second cluster to obtain target obstacle information corresponding to each second cluster, thereby obtaining multiple target obstacle information.

[0149] In some embodiments, the device may further include: The first conversion module is used to convert the information of each target obstacle into target visualization tags to obtain multiple target visualization tags; The first publishing module is used to combine all target visualization tags into a target tag array message and publish the target tag array message so that visualization tools can visualize and display the target tag array message.

[0150] In some embodiments, the device may further include: The second publishing module is used to encapsulate the minimum target distance into a target rock distance message and publish the rock distance message through the target topic for the upper-level decision control module to subscribe to and use.

[0151] The mine obstacle detection device provided in this application embodiment can be used to execute the mine obstacle detection method, that is, it has the beneficial effects and implementation methods of the mine obstacle detection method provided in this application embodiment. For details, please refer to the specific description of the mine obstacle detection method in the above embodiment, which will not be repeated here.

[0152] Figure 5 This is a block diagram of an electronic device provided in an embodiment of this application.

[0153] Reference Figure 5 This application provides an electronic device, which includes: at least one processor 701; at least one memory 702; and one or more I / O interfaces 703 connected between the processor 701 and the memory 702; wherein the memory 702 stores one or more computer programs that can be executed by at least one processor 701, and the one or more computer programs are executed by at least one processor 701 to enable at least one processor 701 to perform the above-described obstacle detection method in the mining area.

[0154] This application also provides a computer-readable storage medium storing a computer program thereon, wherein the computer program, when executed by a processor / processor core, implements the above-described method for detecting obstacles in a mining area. The computer-readable storage medium can be volatile or non-volatile.

[0155] This application also provides a computer program product, including computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code. When the computer-readable code is run in the processor of an electronic device, the processor in the electronic device executes the above-described mining obstacle detection method.

[0156] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0157] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0158] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0159] The computer program instructions used to perform the operations of this application may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuits, such as programmable logic circuits, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), are personalized by utilizing the status information of the computer-readable program instructions. These electronic circuits can execute the computer-readable program instructions to implement various aspects of this application.

[0160] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0161] Various aspects of this application are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0162] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0163] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0164] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0165] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for general illustrative purposes only and should not be construed as limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of this application as set forth by the appended claims.

Claims

1. A method for detecting obstacles in a mining area, characterized in that, include: Obtain the first point cloud data; The first point cloud data is segmented into ground data to obtain non-ground point cloud data. The non-ground point cloud data is height filtered to obtain second point cloud data; the first height value corresponding to each first point cloud in the second point cloud data is within a preset height range. Clustering the second point cloud data yields multiple first clusters; Feature analysis is performed on each of the first clusters to obtain target obstacle information corresponding to each of the first clusters, so as to obtain multiple target obstacle information; The target obstacle information includes at least one of the following: the target obstacle's center position, target size, and target distance; The target distance is used to characterize the distance between the target obstacle and the nearest point of the target vehicle.

2. The method according to claim 1, characterized in that, The acquisition of the first point cloud data includes: Acquire lidar point cloud data; The lidar point cloud data is preprocessed to obtain the first point cloud data; the preprocessing includes at least one of format conversion processing and voxel filtering downsampling processing.

3. The method according to claim 1, characterized in that, The step of performing ground segmentation on the first point cloud data to obtain non-ground point cloud data includes: Obtain the second altitude value corresponding to each second point cloud in the first point cloud data; Multiple third point clouds are selected from all the second point clouds; the third height value corresponding to the third point cloud is greater than or equal to the sum of a preset distance threshold and the minimum second height value; All of the aforementioned third point clouds are identified as non-ground point cloud data.

4. The method according to claim 1, characterized in that, The step of performing ground segmentation on the first point cloud data to obtain non-ground point cloud data includes: The first point cloud data is segmented into ground data using a random sampling consensus algorithm to obtain the non-ground point cloud data.

5. The method according to claim 1, characterized in that, The clustering of the second point cloud data yields multiple first clusters, including: A density-based spatial clustering algorithm with noise is used to cluster the second point cloud data to obtain multiple first clusters.

6. The method according to claim 1, characterized in that, Before performing feature analysis on each of the first clusters to obtain target obstacle information corresponding to each of the first clusters, and thus obtaining multiple sets of target obstacle information, the method further includes: Based on a preset range of points, the first cluster is filtered to obtain a second cluster; the number of points in the second cluster is within the preset range. The feature analysis is performed on each of the first clusters to obtain target obstacle information corresponding to each of the first clusters, thereby obtaining multiple sets of target obstacle information, including: Feature analysis is performed on each of the second clusters to obtain target obstacle information corresponding to each of the second clusters, thereby obtaining multiple target obstacle information.

7. The method according to claim 1, characterized in that, After performing feature analysis on each of the first clusters to obtain target obstacle information corresponding to each of the first clusters, and obtaining multiple sets of target obstacle information, the method further includes: For each of the target obstacle information, it is converted into target visualization tags to obtain multiple target visualization tags; All the target visualization tags are combined into a target tag array message, and the target tag array message is published for visualization tools to visualize.

8. The method according to claim 1, characterized in that, After performing feature analysis on each of the first clusters to obtain target obstacle information corresponding to each of the first clusters, and obtaining multiple sets of target obstacle information, the method further includes: The minimum target distance is encapsulated into a target rock distance message, and the rock distance message is published through the target topic for the upper-level decision control module to subscribe to and use.

9. A mine obstacle detection device, characterized in that, include: The first acquisition module is used to acquire the first point cloud data; The first segmentation module is used to perform ground segmentation on the first point cloud data to obtain non-ground point cloud data. The first filtering module is used to perform height filtering on the non-ground point cloud data to obtain second point cloud data; the first height value corresponding to each first point cloud in the second point cloud data is within a preset height range. The first clustering module is used to cluster the second point cloud data to obtain multiple first clusters; The first analysis module is used to perform feature analysis on each of the first clusters to obtain target obstacle information corresponding to each of the first clusters, so as to obtain multiple target obstacle information. The target obstacle information includes at least one of the following: the target obstacle's center position, target size, and target distance; The target distance is used to characterize the distance between the target obstacle and the nearest point of the target vehicle.

10. 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 one or more computer programs that can be executed by the at least one processor, the one or more computer programs being executed by the at least one processor to enable the at least one processor to perform the mining obstacle detection method as described in any one of claims 1-8.

Images