A uniform rolling method for mine area automatic driving vehicle

Abstract

The present application relates to a kind of uniform rolling method for mine area automatic driving vehicle, belong to the technical field of agent automation driving control, to solve the problem that existing automatic driving vehicle cannot accurately obtain the information of rolling area when running in mine area, and the same rolling area is repeatedly rolled over.The present application plans the driving path of automatic driving vehicle by obtaining the information of rolling area in mine area, so that the same road surface is rolled over with uniform times when automatic driving vehicle passes through the road surface of driving path, realizes that when automatic driving vehicle repeatedly drives between the road of loading area and unloading area in mine area, it will not be concentrated in some positions repeatedly rolled over on a road area in width direction, and the original rolling area is rolled flat or not repeatedly rolled over the same rolling area to make it produce greater protrusion.

Description

Technical Field

[0001] This invention belongs to the field of intelligent agent automated driving control technology, and relates to a uniform compaction method for autonomous vehicles in mining areas. Background Technology

[0002] Due to the unique geographical environment of open-pit mining areas, the overall hardness of the soil and rock layers on the coal surface and coal seam is relatively low. Especially during the freezing and thawing of underground soil in spring and the rainy season, the road surface hardness is insufficient to support the repeated passage and compaction of various mining trucks, which will cause very obvious road ruts. In severe cases, the raised ruts will cause vehicles to bottom out.

[0003] Because the perception field of sensors on autonomous vehicles is limited, the environmental information obtained by the vehicles is incomplete. This leads to inaccurate path planning when autonomous vehicles travel on roads with ruts, posing safety hazards. To enrich the information acquired by the vehicles, intelligent road devices can be installed near the roads. These devices can collect dynamic and static information about the mining area road environment and participants in the mining transportation scenario, and then send the collected information to a cloud server. The information collected by the vehicles is also sent to the cloud server. The cloud server integrates the information from various aspects to form a cloud control platform. The cloud control platform integrates the acquired information and then synchronizes it to the vehicles, enabling the vehicles to fuse the environmental information they perceive and the received information, and perform pre-planned global path planning based on this. However, when autonomous mining vehicles travel on soft, unstructured road surfaces based on this pre-planned global path, random bumps such as rutting and slush can occur. If the same rutting area is repeatedly driven over, the existing bumps will be aggravated. Autonomous vehicles that subsequently arrive at the same rutting area cannot avoid, detour, or ride over it. This not only reduces transportation efficiency but also poses a significant safety hazard to the operation of autonomous mining vehicles. Summary of the Invention

[0004] In view of the above problems, the present invention provides a uniform compaction method for autonomous vehicles in mining areas, which solves the problem that existing autonomous vehicles cannot accurately obtain information about the frost heave area when operating in mining areas, and repeatedly drive over and compact the same frost heave area.

[0005] This invention provides a uniform compaction method for autonomous vehicles in mining areas, specifically including the following steps: Preliminary global path planning yields a preliminary global path; Global uniform path planning yields a globally uniform path: Construct a point cloud prior map of the driving area in the mining area; The road boundary distance is obtained based on the point cloud prior map; By combining the initial global path, road boundary distance, and autonomous vehicle size, the number of uniform compaction passes of the autonomous vehicle in a section of road in the road width direction is adjusted to obtain a global uniform path. The expression for the number of uniform compaction passes is: d = (nm) / w; Where n is the distance to the road boundary; m is the width of the autonomous vehicle; w is the total width of the tires on one side of the axle of the autonomous vehicle; d is the number of uniform rolling passes between the two boundaries of a section of road in the road width direction; the autonomous vehicle adjusts the lateral translation distance when it moves from one side to the other after each rolling pass when it travels in one end of the road area in the road width direction according to the number of uniform rolling passes d, so as to obtain a globally uniform path; Obtain the frost heave area of ​​the driving area of ​​the autonomous vehicle in real time: The autonomous vehicle travels in real time along a globally uniform path to obtain the current frame point cloud. The current frame point cloud is then compared with the point cloud data of the scene corresponding to the current frame image in the prior map of the mining area. If there is a frost heave area, the point cloud of the frost heave area and the point cloud of the flat ground area are segmented. The real-time driving path of autonomous vehicles is locally adjusted based on the point cloud of the segmented frost heave area and the point cloud of the flat ground area.

[0006] Optionally, the specific steps for obtaining the road boundary distance based on the point cloud prior map are as follows: Obtain the training point cloud dataset of the mining area, label the road boundaries in the training point cloud dataset, and train the labeled road boundary point cloud using deep learning to obtain the road boundary weight model; The prior point cloud map is input into the road boundary weight model to obtain the road boundary distance.

[0007] Optionally, by combining the current location of the autonomous vehicle with the segmented point cloud of the frost heave area and the point cloud of the flat ground area, multiple point clouds that are closest to the current frame point cloud in terms of timestamp are selected and overlaid to obtain detailed point cloud data of the frost heave area and detailed point cloud data of the flat ground area. The real-time driving path of autonomous vehicles is locally adjusted based on the point cloud of the frost-prone area and the point cloud of the flat ground area.

[0008] Optionally, the current position of the autonomous vehicle can be obtained using the ICP point cloud localization method with map representation, based on the point cloud prior map and the current frame point cloud.

[0009] Optionally, the specific steps for partially adjusting the real-time driving path of an autonomous vehicle are as follows: Calculate the height and area of ​​the point cloud data for the frost heave region; The acquired frost heave region with point cloud data is divided into meshes, and the frost heave region is divided into... n The side length is m The grid; the average height of a single grid is calculated based on point cloud data of all point cloud data within each grid, using data from flat ground areas. K 1, K 2,…, K n and the overall average height of all point cloud data within the entire frost-affected area. L The average height difference of each grid is obtained based on the average height of a single grid cell and the overall average height. H i , H i =| LK i |, K i For the first i The average height of a single grid cell. i ∈1,2,…, n ; Calculate the average height difference of point clouds between two adjacent grids. h , h =| K i - K j |, K j For the first i The average height of a single grid cell in adjacent grid cells. j ∈1,2,…, n If the average height difference of the point clouds of two adjacent grids h Less than the average height difference between the two adjacent grids H i and H j If the average value is used, then the point cloud data within the two adjacent grids is recorded; traversing the point cloud height differences between all two adjacent grids in the frost heave area, the highest height value among all recorded point cloud data is selected as the maximum height of the frost heave area. H ; The area of ​​the heaving zone is obtained based on the number and size of the grid. A ; Based on the height and area of ​​the frost heave zone, the real-time driving path of the autonomous vehicle is locally adjusted.

[0010] Optionally, the specific steps for making partial adjustments to the real-time driving path of autonomous vehicles are as follows: If the maximum height of the frost heave area H < L Then the autonomous vehicle travels along a globally uniform path; if the maximum height of the frost heave area...H > L Based on the area of ​​the frost heave zone A The size of the vehicle's path was adjusted to avoid the frost-prone area; among other things... L The maximum height of the frost-affected area that would impede the normal operation of autonomous vehicles.

[0011] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects: (1) The present invention uses point cloud segmentation to segment the road boundary of the preliminary global path planning obtained in advance, and then uses the road width information obtained from the left and right boundaries of the road obtained from the prior map, and combines the tire width to adjust the translation distance of the autonomous vehicle when driving back and forth on different widths, so as to realize the global path planning optimization of the vehicle.

[0012] (2) The present invention adjusts the real-time path planning of autonomous vehicles in real time based on the ground frost heave area detected by point cloud background difference and the geometric size information of the frost heave area; if there is no frost heave, the vehicle will drive along the original path; if there is frost heave, the path planning will be adaptively adjusted according to the height of the frost heave area to allow the autonomous vehicle to bypass the frost heave area or directly run over and pass through the frost heave area to achieve the purpose of local path optimization adjustment.

[0013] (3) This invention combines global path planning optimization and local path adjustment optimization to adjust the autonomous vehicle in real time. By obtaining the number of trips of the global path, the decision on whether to compact the road surface is made based on the height of the frost heave area obtained after point cloud background difference processing. The remaining compaction trips are calculated based on the distance between the non-frost heave area and the boundary direction of the road and fed back to the global path planning. This provides a global uniform compaction path reference for the vehicle and the following vehicles. Finally, the purpose of uniform compaction is achieved through global path planning and local path adjustment. Attached Figure Description

[0014] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0015] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a diagram showing the effect of ground frost heave after differential analysis of real-time lidar point cloud and prior map according to the present invention. Figure 3 This is a diagram showing the state of road surface frost heave after applying the method of the present invention.

[0016] Figure label: A - Prior map point cloud; B - Current frame point cloud of the vehicle in real time; C - Point cloud of the frost heave area obtained after background subtraction. Detailed Implementation

[0017] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0018] This invention plans the driving path of an autonomous vehicle by acquiring information about the frost heave area in the mining area. When the autonomous vehicle passes through the road surface of the driving path, it rolls over the same road surface a uniform number of times. This ensures that when the autonomous vehicle repeatedly drives between the loading and unloading areas of the mining area, it will not repeatedly drive over and roll over certain locations in a section of the road in the width direction. This will either flatten the original frost heave area or avoid repeatedly rolling over the same frost heave area to prevent it from becoming larger.

[0019] A specific embodiment of the present invention, such as Figure 1-3 A uniform compaction method for autonomous vehicles in mining areas is disclosed, which specifically includes the following steps: S1: Preliminary global path planning; Acquire the starting position information, GNSS / IMU, and lidar point cloud data of the autonomous vehicle during the initial operation phase; and perform preliminary global path planning based on the starting position information and the operating environment of the autonomous vehicle to obtain a preliminary global path.

[0020] Optionally, the location information of the autonomous vehicle can be obtained through a positioning base station and a C-V2X communication system; the lidar is installed at the front end of the autonomous vehicle.

[0021] Optionally, the operating environment of the autonomous vehicle includes the entire driving distance, road boundaries, and initial road surface frost conditions.

[0022] It is understood that the road boundary described in this invention refers to the boundary of the autonomous vehicle in the width direction of the road it is driving on.

[0023] S2: Global Uniform Path Planning; S2-1: Establish a 3D point cloud map based on point cloud data from GNSS / IMU and LiDAR, and construct a priori point cloud map of the driving area in the mining area by combining the unstructured scene characteristics of the mining area.

[0024] Optionally, the unstructured scene features include the planar features of the road surface and the marking features such as signposts and markers set up in the unstructured scene.

[0025] S2-2: Obtaining road boundary distances based on point cloud prior maps; Obtain the training point cloud dataset of the mining area, label the road boundaries in the training point cloud dataset, and train the labeled road boundary point cloud using deep learning to obtain the road boundary weight model; The road boundary distance is obtained by inputting the prior point cloud map into the road boundary weight model.

[0026] S2-3: Combine the initial global path, road boundary distance, and total tire width on one side of the axle to adjust the number of uniform rolling passes of the autonomous vehicle in a section of road in the road width direction to obtain a global uniform path. Specifically, the road boundary distance is n; the vehicle width is m; and the total width of the tires on one side of the axle is w. In the initial global path, the autonomous vehicle travels from one side to the other between the two boundaries in the road width direction for one cycle, and obtains the number of trips d for one cycle, d = (nm) / w. Based on the number of trips d, the autonomous vehicle adjusts the lateral translation distance when performing the next compaction after each compaction in the road width direction to obtain a globally uniform path. This ensures that when the same autonomous vehicle travels within the same road surface area in the road width direction, it does not repeatedly travel to the same position in that area, preventing the same autonomous vehicle from repeatedly driving through and compacting the road surface at the same location, which could lead to increased rutting and the formation of a larger sludge heave area. This achieves the goal of the vehicle traveling along a uniform path in the initial operation phase.

[0027] S3: Obtain the frost heave area of ​​the driving area during real-time operation; The autonomous vehicle travels in real time along a globally uniform path to acquire the current frame's point cloud. It then performs background subtraction processing on the current frame's point cloud and the point cloud data of the scene corresponding to the current frame image in the prior point cloud map constructed in step S2-1. If a frost-affected area exists, the point cloud of the frost-affected area and the point cloud of the flat ground area are segmented, and the process proceeds to the next step. If no frost-affected area exists, the autonomous vehicle continues traveling along the globally uniform path. (See also...) Figure 2 The image shows the effect of ground frost heave after differential analysis of real-time lidar point cloud and prior map. S4: Overlay of multiple point cloud data frames; Based on the point cloud prior map and the current frame point cloud, the ICP point cloud localization method with map representation is used to obtain the current position of the autonomous vehicle. By combining the location information of the autonomous vehicle with the point cloud of the frost heave area and the point cloud of the flat ground area segmented in step S3, and selecting the point cloud of multiple frames that are closest to the current frame in terms of timestamp, we can obtain the detailed point cloud data of the frost heave area and the detailed point cloud data of the flat ground area.

[0028] Preferably, the three most recent consecutive frames of point cloud data are selected for overlay.

[0029] S5: Local path adjustment; S5-1: Calculate the height and area of ​​the point cloud data for the frost heave region; The meticulous frost heave region with point cloud data obtained in step S4 is divided into meshes, and the meticulous frost heave region is divided into... n The side length is m The grid; the average height of a single grid is calculated based on point cloud data of a meticulously flat ground area, representing all point cloud data within that grid. K 1, K 2,…, K n And the overall average height of all point cloud data within the entire meticulous frost heave area. L The average height difference of each grid is obtained based on the average height of a single grid cell and the overall average height. H i , H i =| LK i |, K i For the first i The average height of a single grid cell. i ∈1,2,…, n ; Calculate the average height difference of point clouds between two adjacent grids. h , h =| K i - K j |, K j For the first i The average height of a single grid cell in adjacent grid cells. j ∈1,2,…, n The original point cloud is selected based on the following conditions: if the average height difference between two adjacent grid points is less than the average height difference between the point clouds... h Less than the average height difference between the two adjacent grids H i and H j If the average height difference between the two adjacent grid points is the mean, then the original point cloud within those two adjacent grids is used; if the average height difference between the point clouds of the two adjacent grids is less than the mean height difference, then the original point cloud within those two adjacent grids is used. h If the height difference is greater than the average of the height differences between adjacent grids, it is not accepted. The height difference between all two adjacent grids in the frost heave area is traversed, and all accepted point cloud data are recorded. The height value of the highest point among all accepted point cloud data is selected as the maximum height of the frost heave area. H .

[0030] The area of ​​the meticulous frost heave zone is obtained based on the number and size of the grid. A That is, the number of grids in the frost heave area is n, and the area of ​​the frost heave area is... A for n × m × m .

[0031] Preferably, the side length of the grid is 0.3 × 0.3 m.

[0032] S5-2: Based on the height and area of ​​the carefully measured frost heave zone, make local adjustments to the real-time driving path of the autonomous vehicle; Specifically, the maximum height of the frost heave zone that affects the normal operation of autonomous vehicles is [missing information]. L If the maximum height of the frost heave area H < L Then the autonomous vehicle travels along a globally uniformly planned path; if the maximum height of the frost heave area... H > L Based on the area of ​​the frost heave zone A The size of the vehicle affects the local path adjustment, allowing it to bypass the frost-covered area.

[0033] In summary, by employing the globally uniform path planning method developed in this invention and combining it with the height and area information of the detected road surface frost heave zone for local path optimization, the goal of achieving reasonable path planning for autonomous vehicles facing frost heave zones during road transportation in mining areas can be ultimately realized. (See [link to relevant documentation]). Figure 3 The invention demonstrates five different ruts obtained by an autonomous vehicle using the method of this invention. Figure 3 The number in the figure represents the initial state 1. Through the method of this invention, the ground effect is achieved by driving in the lateral direction according to the number of times, so as to achieve the purpose of uniform compaction on the road surface in the mining area. Instead of repeatedly compacting in one position, the road surface is free from obvious road heave and bulge.

[0034] It is understood that the present invention uses an autonomous mining vehicle, which is equipped with an RTK GNSS / IMU integrated inertial navigation system, a C-V2X communication system and a lidar. Based on the installed RTK GNSS / IMU integrated inertial navigation system, C-V2X communication system and lidar, the driving path planning in the muddy area of ​​the mining area is carried out.

[0035] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A uniform compaction method for autonomous vehicles in mining areas, characterized in that, Specifically, the following steps are included: Preliminary global path planning yields a preliminary global path; Global uniform path planning yields a globally uniform path: Construct a point cloud prior map of the driving area in the mining area; The road boundary distance is obtained based on the point cloud prior map; By combining the initial global path, road boundary distance, and autonomous vehicle size, the number of uniform compaction passes of the autonomous vehicle in a section of road in the road width direction is adjusted to obtain a global uniform path. The expression for uniform compaction passes is: d = (nm) / w; Where n is the distance to the road boundary; m is the width of the autonomous vehicle; w is the total width of the tires on one side of the axle of the autonomous vehicle; d is the number of uniform rolling passes between the two boundaries of a section of road in the road width direction; the autonomous vehicle adjusts the lateral translation distance when it moves from one side to the other after each rolling pass when it travels in one end of the road area in the road width direction according to the number of uniform rolling passes d, so as to obtain a globally uniform path; Obtain the frost heave area of ​​the driving area of ​​the autonomous vehicle in real time: The autonomous vehicle travels in real time along a globally uniform path to obtain the current frame point cloud, and performs background subtraction processing on the current frame point cloud and the point cloud data of the scene corresponding to the current frame image in the prior map of the point cloud of the driving area in the mining area. If there is a frost heave area, separate the point cloud of the frost heave area and the point cloud of the flat ground area; The real-time driving path of autonomous vehicles is locally adjusted based on the point cloud of the frost heave area and the point cloud of the flat ground area.

2. The uniform compaction method according to claim 1, characterized in that, The specific steps for obtaining road boundary distances based on the aforementioned point cloud prior map are as follows: Obtain the training point cloud dataset of the mining area, label the road boundaries in the training point cloud dataset, and train the labeled road boundary point cloud using deep learning to obtain the road boundary weight model; The prior point cloud map is input into the road boundary weight model to obtain the road boundary distance.

3. The uniform compaction method according to claim 1, characterized in that, By combining the current location of the autonomous vehicle with the segmented point cloud of the frost heave area and the point cloud of the flat ground area, multiple point clouds that are closest to the current frame point cloud in terms of timestamp are selected and overlaid to obtain detailed point cloud data of the frost heave area and detailed point cloud data of the flat ground area. The real-time driving path of autonomous vehicles is locally adjusted based on the point cloud of the frost-prone area and the point cloud of the flat ground area.

4. The uniform compaction method according to claim 3, characterized in that, Based on the point cloud prior map and the current frame point cloud, the current position of the autonomous vehicle is obtained using the ICP point cloud localization method with map representation.

5. The uniform compaction method according to any one of claims 1-4, characterized in that, The specific steps for partially adjusting the real-time driving path of an autonomous vehicle are as follows: Calculate the height and area of ​​the point cloud data for the meticulously prepared frost heaving area; The acquired frost heave region with point cloud data is divided into meshes, and the frost heave region is divided into... n The side length is m The grid; Calculate the average height of a single grid based on point cloud data of all point cloud data within each grid area. K 1, K 2,…, K n and the overall average height of all point cloud data within the entire frost-affected area. L The average height difference of each grid is obtained based on the average height of a single grid cell and the overall average height. H i , H i =| LK i |, K i For the first i The average height of a single grid cell. i ∈1,2,…, n ; Calculate the average height difference of point clouds between two adjacent grids. h , h =| K i - K j |, K j For the first i The average height of a single grid cell in adjacent grid cells. j ∈1,2,…, n If the average height difference of the point clouds of two adjacent grids h Less than the average height difference between the two adjacent grids H i and H j If the average value is used, then the point cloud data within the two adjacent grids is recorded; traversing the point cloud height differences between all two adjacent grids in the frost heave area, the highest height value among all recorded point cloud data is selected as the maximum height of the frost heave area. H ; The area of ​​the heaving zone is obtained based on the number and size of the grid. A ; Based on the height and area of ​​the frost heave zone, the real-time driving path of the autonomous vehicle is locally adjusted.

6. The uniform compaction method according to claim 5, characterized in that, The specific steps for making local adjustments to the real-time driving path of autonomous vehicles are as follows: If the maximum height of the frost heave area H < L Then the autonomous vehicle travels along a globally uniform path; if the maximum height of the frost heave area... H > L Based on the area of ​​the frost heave zone A The size of the vehicle's path was adjusted to avoid the frost-prone area; among other things... L The maximum height of the frost-affected area that would impede the normal operation of autonomous vehicles.

Images