A method, system and device for the parking alignment of terminal container trucks

Abstract

The present invention discloses a method, a system and a device for the parking alignment of terminal container trucks. The method includes obtaining point cloud data of the site where the container truck is to park; obtaining parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data; obtaining model data of the container truck according to the point cloud data, and obtaining contour data of the outer bounding box of the container truck according to the model data; obtaining the container type of the container on the container truck according to the model data of the container truck; determining parameters of the target alignment plane where the head or the tail of the container truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane; obtaining a first distance between the outer bounding box of the container truck and the target alignment plane, and when the first distance is less than or equal to a preset distance, determining that the parking alignment of the container truck is successful, so as to realize the automation of the parking alignment of terminal container trucks.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of quay container truck parking alignment, and in particular, to a quay container truck parking alignment method, system and device. Background Art

[0002] In a port, during the container loading and unloading operation of container trucks, many mechanical devices need to perform long-term repetitive work. If operated manually, it will waste a lot of human resources and result in low work efficiency. To improve efficiency and solve potential safety hazards, many ports have started to explore how to transform old port loading and unloading equipment. Summary of the Invention

[0003] The present invention provides a quay container truck parking alignment method, system and device to achieve the automation, digitization and intelligence of quay container truck parking alignment.

[0004] To achieve the above object, the first aspect embodiment of the present invention proposes a quay container truck parking alignment method, including the following steps:

[0005] Obtain the point cloud data of the site where the container truck is to park;

[0006] Obtain the parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data;

[0007] Obtain the model data of the container truck according to the point cloud data, and obtain the contour data of the outer bounding box of the container truck according to the model data;

[0008] Obtain the container type of the container truck according to the model data of the container truck;

[0009] Determine the parameters of the target alignment plane where the head or tail of the container truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane;

[0010] Obtain the first distance between the outer bounding box of the container truck and the target alignment plane, and when the first distance is less than or equal to a preset distance, determine that the container truck parking alignment is successful.

[0011] According to an embodiment of the present invention, the obtaining the parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data includes:

[0012] Perform filtering processing on the point cloud data to obtain filtered point cloud data;

[0013] Obtain the parameters of the ground plane of the target parking space according to the filtered point cloud data.

[0014] According to an embodiment of the present invention, obtaining the parameters of the ground plane of the target parking space based on the filtered point cloud data includes:

[0015] Define a segmentation model, and determine whether the current point in the filtered point cloud data is adapted to the segmentation model. If so, classify the current point as an inlier; if not, classify the current point as an outlier;

[0016] Traverse the points in the filtered point cloud data, count the number of all inliers. When the number of inliers is greater than or equal to a preset threshold, update the segmentation model according to all the current inliers, and store all the inliers as the segmentation result;

[0017] When the number of all inliers is less than the preset threshold, determine whether the number of all the current inliers is greater than the number of inliers in the segmentation model. If so, update the segmentation model according to all the current inliers, and store all the inliers as the segmentation result; if not, use the segmentation model as the segmentation result; wherein, the segmentation model is a ground plane model;

[0018] Obtain the parameters of the ground plane of the target parking space according to the segmentation result.

[0019] According to an embodiment of the present invention, before obtaining the parameters of the ground plane of the target parking space based on the filtered point cloud data, it further includes:

[0020] Perform rasterization processing on the filtered point cloud data.

[0021] According to an embodiment of the present invention, obtaining the model data of the container carrier based on the point cloud data and obtaining the contour data of the outer bounding box of the container carrier includes:

[0022] Perform straight-through filtering processing on the point cloud data to obtain straight-through filtered point cloud data;

[0023] Obtain the model data of the container carrier according to the straight-through filtered point cloud data;

[0024] Extract the of the outer bounding box contour data of the container carrier according to the model data.

[0025] According to an embodiment of the present invention, after determining that the container carrier has successfully parked in place, it further includes:

[0026] Model the container according to the point cloud data to obtain the model data of the container;

[0027] Determine whether the container and the truck are successfully aligned according to the model data of the container and the model data of the truck.

[0028] According to an embodiment of the present invention, after the container and the truck are successfully aligned, it further includes:

[0029] Model the spreader through the point cloud data to obtain the model data of the spreader;

[0030] Determine whether the spreader and the container are separated completely according to the model data of the spreader and the model data of the container.

[0031] According to an embodiment of the present invention, after obtaining the point cloud data of the site where the truck is to park, it further includes:

[0032] Perform coloring processing on the point cloud data.

[0033] To achieve the above object, a second aspect embodiment of the present invention proposes a quay truck parking alignment system for executing the quay truck parking alignment method as described above, including:

[0034] A point cloud data acquisition module for acquiring the point cloud data of the site where the truck is to park;

[0035] A ground plane parameter acquisition module for acquiring the parameters of the ground plane of the target parking space where the truck is to park according to the point cloud data;

[0036] A truck model data acquisition module for acquiring the model data of the truck according to the point cloud data and acquiring the contour data of the outer bounding box of the truck according to the model data;

[0037] A container type acquisition module for acquiring the container type of the truck according to the model data of the truck;

[0038] A target alignment plane determination module for determining the parameters of the target alignment plane to which the head or tail of the truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane;

[0039] A judgment module for acquiring the distance between the outer bounding box of the truck and the target alignment plane, and when the distance is less than or equal to a preset distance, determining that the truck parking alignment is successful.

[0040] To achieve the above object, a third aspect embodiment of the present invention proposes a quay truck parking alignment device, including:

[0041] A lidar, a camera, a host computer, a main control module, an industrial computer and a pan-tilt head;

[0042] The host computer is used to send instructions to the main control module, and the main control module forwards the instructions to the industrial control computer. The industrial control computer adjusts the attitude angle of the pan-tilt according to the instructions. The lidar and the camera are arranged on the pan-tilt and are used to scan the point cloud data of the target parking space where the container truck is to park. The lidar and the camera are also used to upload the collected point cloud data to the industrial control computer, and the industrial control computer forwards the point cloud data to the main control module.

[0043] The main control module executes the container truck parking alignment method as described above according to the point cloud data, so that the container truck parks in the target parking space.

[0044] According to an embodiment of the present invention, the container truck parking alignment system further includes: a display screen, which is connected to the main control module; the display screen is used to display the distance between the outer bounding box of the container truck sent by the main control module and the target alignment plane.

[0045] According to the container truck parking alignment method, system and device proposed by the embodiment of the present invention, the method includes obtaining the point cloud data of the site where the container truck is to park; obtaining the parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data; obtaining the model data of the container truck according to the point cloud data, and obtaining the contour data of the outer bounding box of the container truck according to the model data; obtaining the container type of the container truck according to the model data of the container truck; determining the parameters of the target alignment plane to which the head or tail of the container truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane; obtaining the first distance between the outer bounding box of the container truck and the target alignment plane, and when the first distance is less than or equal to the preset distance, it is determined that the container truck parking alignment is successful. Through the above container truck parking alignment method, automatic alignment parking of the container truck at the dock can be realized, and it can also be realized to automatically judge whether the container is aligned with the container truck and whether the spreader is separated from the container. The whole process of container truck parking alignment, container alignment and spreader-container separation at the dock is realized automatically, solving the problem that this process requires manual operation, reducing human-machine contact, eliminating the safety hazards existing in manual operation, ensuring personnel safety, improving operation efficiency, reducing labor costs, and meeting the transformation expectations of port handling equipment. Description of the Drawings

[0046] Figure 1 is a flow chart of the container truck parking alignment method proposed by the embodiment of the present invention;

[0047] Figure 2is a block diagram of a terminal container truck parking alignment device proposed in an embodiment of the present invention;

[0048] Figure 3 It is a block diagram of a terminal container truck parking alignment system proposed in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0050] At present, major ports around the world are changing their previous production methods through digitalization and automation technology. With the rapid development of semi-automated container terminals, the advancement of fully automated container terminal technology has gradually become prominent. However, currently multiple sets of equipment are required to cover the work site, which is not conducive to later upgrades and maintenance. The dock container truck parking alignment method, system and device proposed in the embodiment of the present invention solve the above problems, and also solve the problem of dock container truck parking alignment, and can also automatically determine whether the container and the container truck are aligned, and whether the spreader and the container are separated. By automating the entire process of dock container truck parking alignment, alignment with the container, and separation of the spreader and the container, the problem that the process requires manual operation is solved, human-machine contact is reduced, and the safety hazards of manual operation are eliminated, personnel safety is guaranteed, operating efficiency is improved, and labor costs are reduced, which meets the expectations of the transformation of port loading and unloading equipment.

[0051] Embodiment 1

[0052] Figure 1 1 is a flow chart of the method for parking container trucks at a terminal according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0053] S101, obtaining point cloud data of the parking lot where the container truck is to park;

[0054] It should be noted that the scenarios where the container truck needs to park are as follows: near transportation devices such as gantry cranes in the port terminal. After the container truck parks in the parking area, according to the situation, the gantry crane places the container on the shelf or the gantry tower crane lifts the container from the container truck to complete loading and unloading. Furthermore, the working mechanism of the gantry tower crane determines that it can only perform loading and unloading operations on one container truck at a time. However, there are generally multiple lanes in the operating area. After the container truck enters the area, the staff will indicate which lane the container truck should enter, that is, which specific lane the container truck needs to park in is determined by the staff. Therefore, the first step of the program is to drive the device composed of the lidar and the camera to rotate to the set position according to the lane designated by the staff. The pan-tilt is programmable to control the rotation angle and transmits the real-time attitude angle of the pan-tilt back through the RS422 interface. This attitude angle will participate in the parsing of the scanning data of the lidar and the camera to improve the modeling accuracy. After the pan-tilt drives the scanning device to rotate to the designated position, the pan-tilt will only scan the lane where the container truck parks within a small angle range, and only scan the operating lane of the container truck, which not only ensures the real-time performance of the lidar modeling but also reduces the interference of objects in other lanes on the lidar and improves the accuracy. As the pan-tilt rotates back and forth within the designated angle range, the scanning device composed of the lidar and the camera will synchronously collect data and perform modeling, that is, obtain the point cloud data of the area where the container truck needs to park.

[0055] According to an embodiment of the present invention, after obtaining the point cloud data of the area where the container truck needs to park, it further includes:

[0056] Performing coloring processing on the point cloud data.

[0057] Among them, according to the requirement of real-time performance, this technical solution will perform point cloud modeling at least once within 1 second, and the camera will provide accurate coloring processing for the point cloud color to make the point cloud model more realistic and more convenient for the operator to observe. Among them, through the combination of the camera and the radar, colored point cloud data is formed.

[0058] S102, obtaining the parameters of the ground plane of the target parking space where the container truck needs to park according to the point cloud data;

[0059] According to an embodiment of the present invention, obtaining the parameters of the ground plane of the target parking space where the container truck needs to park according to the point cloud data includes:

[0060] Performing filtering processing on the point cloud data to obtain filtered point cloud data;

[0061] Among them, because there may be some other non-target objects such as empty shelves or traffic cones on the site, it is necessary to perform filtering processing on the just-obtained point cloud to ensure that the obtained point cloud is all ground point cloud.

[0062] Based on the filtered point cloud data, obtain the parameters of the ground plane of the target parking space.

[0063] Among them, the RANSAC (Random Sample Consensus) algorithm can be used to process the filtered point cloud data to obtain the parameters of the ground plane of the target parking space.

[0064] According to an embodiment of the present invention, before obtaining the parameters of the ground plane of the target parking space based on the filtered point cloud data, it further includes:

[0065] Perform voxelization processing on the filtered point cloud data.

[0066] Among them, the scanned site point cloud will be voxelized, and the voxel point cloud of the grid will be used to replace the original point cloud to obtain the downsampled point cloud. This step will improve the point cloud processing speed. The so-called voxelization is to form a three-dimensional voxel grid according to the given point cloud, and use the center point / centroid point of all voxels to approximate the point set contained in the voxel, so as to complete the filtered result obtained by downsampling. Voxelization is a preprocessing step for many point cloud algorithms and can well improve the running speed of the program.

[0067] According to an embodiment of the present invention, obtaining the parameters of the ground plane of the target parking space based on the filtered point cloud data includes:

[0068] Define a segmentation model, and determine whether the current point in the filtered point cloud data is suitable for the segmentation model. If so, classify the current point as an inlier; if not, classify the current point as an outlier;

[0069] Traverse the points in the filtered point cloud data, count the number of all inliers. When the number of inliers is greater than or equal to a preset threshold, update the segmentation model according to all current inliers and store all inliers as the segmentation result;

[0070] When the number of all inliers is less than the preset threshold, determine whether the number of all current inliers is greater than the number of inliers in the segmentation model. If so, update the segmentation model according to all current inliers and store all inliers as the segmentation result; if not, use the segmentation model as the segmentation result; among them, the segmentation model is a ground plane model;

[0071] Obtain the parameters of the ground plane of the target parking space according to the segmentation result.

[0072] Specifically, after obtaining the filtered point cloud, the point cloud segmentation algorithm is started. Here, the RANSAC (Random Sample Consensus) algorithm is used to obtain the planar point cloud. This can directly obtain all the point clouds related to the ground plane because the planar feature of the ground is very obvious, while the planar features of other objects on the site, such as traffic cones, are not so obvious. The so-called segmentation of the Random Sample Consensus algorithm is a process of randomly sampling to eliminate outliers and constructing a basic subset consisting only of inlier data. Its basic idea is: when estimating parameters, instead of treating all possible input data indiscriminately, first design a judgment criterion model (i.e., the segmentation model) for the problem, use this judgment criterion to iteratively eliminate the input data that is inconsistent with the estimated model, and then estimate the model parameters through the correct input data. The point cloud segmentation process based on Random Sample Consensus is as follows: First, randomly select some points from the input point cloud dataset and calculate the parameters of the user-given model (segmentation model). Set a distance threshold for all points in the dataset. If the distance from a point to the model is within the distance threshold, then classify this point as an inlier; otherwise, it is an outlier. Then, count the number of all inliers and determine whether it is greater than the set threshold. If it is, then re-estimate the model with the inliers as the model output and store all the inliers as the segmentation result; if not, compare it with the current maximum number of inliers. If it is greater than the current maximum number of inliers, then replace the current maximum number of inliers and store the current model coefficients, and then perform iterative calculations until a satisfactory segmentation result is obtained. Because the features of the ground are very obvious, the segmentation model set in the present invention is a planar model. And because the number of ground point clouds accounts for the majority in the obtained point clouds, generally, setting a certain number of iterations can easily obtain the result. After obtaining the ground point cloud, it is very easy to obtain the planar parameters of the ground. Since the process of obtaining the planar coefficients involves the processes of ground fitting and vector calculation, the real-time performance is poor. Fortunately, this step only needs to be performed once after the gantry tower crane changes the site to obtain the ground plane model here, and this step will not be executed again during the subsequent program operation.

[0073] S103. Obtain the model data of the container truck according to the point cloud data, and obtain the contour data of the outer bounding box of the container truck according to the model data;

[0074] According to an embodiment of the present invention, obtaining the model data of the container truck according to the point cloud data and obtaining the contour data of the outer bounding box of the container truck according to the model data includes:

[0075] Perform a pass-through filter processing on the point cloud data to obtain the pass-through filtered point cloud data. That is to say, not all the collected point cloud data is valid data. For example, the ground point cloud is the part that needs to be removed. Since the ground model parameters of the work site have been obtained previously, according to this plane parameter, the program directly uses the pass-through filter method to filter out the ground points. The so-called pass-through filter is to limit a certain field of the point cloud. For example, filter out all the point clouds whose X-axis is not within the limited field range in the point cloud.

[0076] Obtain the model data of the truck according to the pass-through filtered point cloud data;

[0077] Extract the of the outer bounding box contour data of the truck according to the model data.

[0078] Among them, the Euclidean clustering algorithm can be used to obtain the model of the truck and remove the discrete points. For Euclidean clustering, the distance judgment criterion is the Euclidean distance. For a certain point P in space, find k points closest to point p through the KD-Tree nearest neighbor search algorithm. Those points with a distance less than the set threshold are clustered into the set Q. If the number of elements in Q no longer increases, the entire clustering process ends; otherwise, a point other than point p must be selected in the set Q, and the above process is repeated until the number of elements in Q no longer increases. Here, the Euclidean clustering based on the CPU is adopted, and it will consume CPU resources during operation. Among them, when actually looking for the point cloud data of the truck, the center point of the truck can be used as point P, and the length, width, and height of the truck itself can be used as the set threshold to find the clustering points of the truck.

[0079] It should be noted that the KD-Tree nearest neighbor search algorithm here is an abbreviation of K-dimension tree, which is a data structure for partitioning data points in a k-dimensional space. This is a data structure used in computer science to organize and represent a set of points in a k-dimensional space. It is a binary search tree with other constraints. The kd-tree is very useful for range and nearest neighbor searches. In practice, it is generally three-dimensional, so it will only be processed in three dimensions. Therefore, all kd-trees will be three-dimensional kd-trees. In theory, each level of the kd-tree separates all child nodes in the specified dimension. At the root of the tree, all child nodes are separated in the first specified dimension (that is, if the point with the first coordinate is less than the root node, it will be in the left subtree, and the point greater than the root node will be in the right subtree). Each level of the tree is separated in the next dimension, and after all other dimensions are used up, it returns to the first dimension. The most efficient way to build a kd-tree is to use the partitioning method like quicksort, placing the value of the specified dimension at the root, with the smaller values in the left subtree and the larger values in the right subtree in that dimension. Then repeat this process on the left and right subtrees until the last tree to be sorted consists of only one element.

[0080] After obtaining the set of card point cloud model data, iterate these data to obtain key data such as the center point, the farthest points and the nearest points in each direction. These data are the contour data of the outer bounding box of the model, which on the one hand provides data support for distance measurement and on the other hand provides parameters for visualization.

[0081] S104, obtain the container type of the truck according to the model data of the truck;

[0082] It should be noted that after obtaining the model data of the truck, the container type of the truck will be obtained. Among them, the corresponding relationship between the model data of the truck and the container type corresponding to the truck can be pre-stored in the controller.

[0083] S105, determine the parameters of the target alignment plane where the head or tail of the truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane;

[0084] After obtaining the container type of the current truck at step S104, it is possible to determine the specific parking position of the truck at the target parking space. For example, the distances from the front stop line and the rear stop line, etc. These distances can be pre-stored in the controller in advance. For example, if the distance from the front stop line is obtained, then based on the parameters of the ground plane of the target parking space and this distance, a target alignment plane perpendicular to the ground plane of the target parking space can be obtained. When the front of the truck coincides with this target alignment plane or the distance is within a certain range, the truck's parking alignment is successful. Or, for example, if the distance from the rear stop line is obtained, then based on the parameters of the ground plane of the target parking space and this distance, a target alignment plane perpendicular to the ground plane of the target parking space can be obtained. When the rear of the truck coincides with this target alignment plane or the distance is within a certain range, the truck's parking alignment is successful.

[0085] S106. Obtain the first distance between the outer bounding box of the truck and the target alignment plane. When the first distance is less than or equal to the preset distance, it is determined that the truck's parking alignment is successful.

[0086] That is to say, using the ground plane parameters obtained at the beginning, a plane parameter perpendicular to the ground is obtained. With this plane parameter perpendicular to the ground and the installation position relationship of the scanning device on the gantry crane, a plane perpendicular to the ground at the parking space can be obtained. According to the point cloud outer bounding box data obtained from processing the truck model data before, the program will identify the container type loaded on the truck and adjust the position of the plane accordingly. Then, calculating the distance from the truck to this plane is the distance from the truck to the parking space. And since the program has just obtained the contour data of the outer bounding box of the truck, it is easy to obtain the distance from the truck to the designated parking space according to the distance formula from a point to a plane. This distance value will be displayed in real time on the LED display screen set beside to guide the driver to park the truck at the designated position.

[0087] According to an embodiment of the present invention, after determining that the truck's parking alignment is successful, it further includes:

[0088] Model the container based on the point cloud data to obtain the model data of the container;

[0089] According to the model data of the container and the model data of the truck, determine whether the container and the truck are aligned successfully.

[0090] It can be understood that if the container truck is performing a loading operation, it is also necessary to model the container based on the point cloud data to obtain the model data of the container. The method for obtaining the model data of the container can refer to the method for obtaining the model data of the container truck, that is, the Euclidean clustering method. That is, when the container truck stops stably, the program will model the vehicle and the container under the spreader respectively according to the height difference, obtain the horizontal distance between the container and the truck head according to the point cloud coordinates, analyze whether the container will hit the truck head according to the current falling trend to prevent the container from hitting the truck head, or obtain the distance between the container and the truck tail, analyze whether the container will hit the truck tail according to the current falling trend to prevent the container from hitting the truck tail, or obtain the distance between the container and the edge of the vehicle side to determine whether the container is stably placed on the container truck.

[0091] According to an embodiment of the present invention, after the container and the container truck are successfully aligned, it further includes:

[0092] Model the spreader through the point cloud data to obtain the model data of the spreader;

[0093] According to the model data of the spreader and the model data of the container, determine whether the separation between the spreader and the container is completed.

[0094] It can be understood that when the spreader completes the operation, by analyzing the point cloud model, it can be determined whether the separation between the spreader and the container or between the container and the container truck is completed, which can effectively prevent the container truck from dragging the spreader. Among them, the method for obtaining the model data of the spreader can refer to the method for obtaining the model data of the container truck, that is, the Euclidean clustering method.

[0095] Based on this, the technical solution realizes the automation, digitization and intelligence of the container truck parking alignment at the terminal, and realizes the automation, digitization and intelligence of the container loading and unloading of the container truck at the terminal, greatly improving the efficiency of the container truck operation at the terminal and enhancing the accuracy of the container truck parking, providing convenience for the staff and the container truck drivers.

[0096] Embodiment 2

[0097] Figure 2 It is a schematic block diagram of the container truck parking alignment device proposed in the embodiment of the present invention. As Figure 2 shown, the device 100 is used to execute the previous container truck parking alignment method, including:

[0098] The point cloud data acquisition module 101 is used to acquire the point cloud data of the site where the container truck is to park;

[0099] The ground plane parameter acquisition module 102 is used to acquire the parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data;

[0100] The truck model data acquisition module 103 is configured to obtain the model data of the truck according to the point cloud data, and obtain the contour data of the outer bounding box of the truck according to the model data;

[0101] The container type acquisition module 104 is configured to obtain the container type of the truck according to the model data of the truck;

[0102] The target alignment plane determination module 105 is configured to determine the parameters of the target alignment plane to which the front or rear of the truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane.

[0103] The judgment module 106 is configured to obtain the distance between the outer bounding box of the truck and the target alignment plane, and judge that the truck parking alignment is successful when the distance is less than or equal to a preset distance.

[0104] According to an embodiment of the present invention, the ground plane parameter acquisition module includes:

[0105] A filtering unit for filtering the point cloud data to obtain filtered point cloud data;

[0106] A processing unit for obtaining the parameters of the ground plane of the target parking space according to the filtered point cloud data.

[0107] According to an embodiment of the present invention, the algorithm processing unit is configured to,

[0108] Define a segmentation model, and judge whether the current point in the filtered point cloud data adapts to the segmentation model. If so, classify the current point as an inlier; if not, classify the current point as an outlier;

[0109] Traverse the points in the filtered point cloud data, count the number of all inliers. When the number of inliers is greater than or equal to a preset threshold, update the segmentation model with all current inliers and store all inliers as the segmentation result;

[0110] When the number of all inliers is less than the preset threshold, judge whether the number of all current inliers is greater than the number of inliers in the initial segmentation model. If so, update the segmentation model with all current inliers and store all inliers as the segmentation result; if not, use the segmentation model as the segmentation result; wherein, the segmentation model is a ground plane model;

[0111] Obtain the parameters of the ground plane of the target parking space according to the segmentation result.

[0112] According to an embodiment of the present invention, the device further includes: a coloring module for coloring the point cloud data.

[0113] According to an embodiment of the present invention, the device further includes:

[0114] A rasterization module for rasterizing filtered point cloud data.

[0115] According to an embodiment of the present invention, the truck model data acquisition module includes:

[0116] A direct-pass filtering processing unit for performing direct-pass filtering on point cloud data to obtain direct-pass filtered point cloud data;

[0117] An algorithm unit for obtaining the model data of the truck according to the direct-pass filtered point cloud data;

[0118] A contour data extraction unit for extracting the of the outer bounding box contour data of the truck according to the model data.

[0119] According to an embodiment of the present invention, it further includes:

[0120] A container model data acquisition module for modeling the container according to the point cloud data to obtain the model data of the container;

[0121] A second judgment module for judging whether the container and the truck are successfully aligned according to the model data of the container and the model data of the truck.

[0122] According to an embodiment of the present invention, it further includes:

[0123] A spreader model data acquisition module for modeling the spreader through the point cloud data to obtain the model data of the spreader;

[0124] A third judgment module for judging whether the spreader and the container are separated completely according to the model data of the spreader and the model data of the container.

[0125] The above product can execute the method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the executed method. The detailed content has been explained in the method embodiment and will not be repeated here.

[0126] Embodiment III

[0127] Figure 3 It is a schematic block diagram of a terminal truck parking alignment system proposed by an embodiment of the present invention. As Figure 3 shown, the system 200 includes:

[0128] A lidar 201, a camera 202, a host computer 203, a main control module 204, an industrial computer 205, and a pan-tilt 206;

[0129] The host computer 203 is used to send instructions to the main control module 204, and the main control module 204 forwards the instructions to the industrial control computer 205. The industrial control computer 205 adjusts the attitude angle of the pan-tilt 206 according to the instructions. The lidar 201 and the camera 202 are arranged on the pan-tilt 206 and are used to scan the point cloud data of the target parking space where the container truck is to park. The lidar 201 and the camera 202 are also used to upload the collected point cloud data to the industrial control computer 205, and the industrial control computer 205 forwards the point cloud data to the main control module 204.

[0130] The main control module 204 executes the container truck parking alignment method as described above according to the point cloud data, so that the container truck parks in the target parking space.

[0131] It should be noted that the main control module 204 can be a PLC controller in the port, which is the location of the control center. The staff in the control room will send lane information to the main control module 204 through the console (host computer 203). Then the main control module 204 will forward it to the industrial control computer 205. Then the industrial control computer 205 will send the message to the container truck alignment control software deployed therein to control the pan-tilt 206 to rotate to the pre-set angle, so that the pan-tilt 206 drives the scanning device to rotate to the angle corresponding to the scanning lane. At the same time, the pan-tilt 206 will also transmit the current attitude angle of the pan-tilt in real time. This attitude angle will be sent to the main control module 204 together with the scanning device data for data analysis of modeling, and then the correct instructions will be issued and conveyed to the container truck driver through the LED display screen. Among them, the pan-tilt 206 with a programmable rotation angle is equipped with a scanning device composed of a 2D lidar 201 and a camera 202, which is installed in the middle part at the bottom of the quay crane. The pan-tilt 206 drives the scanning device to scan the specified lane at the specified angle. Currently, there are generally 6 lanes in the port parking lot, and one vehicle is undergoing loading and unloading operations at a time among the 6 lanes. Therefore, as long as the pan-tilt 206 is rotated to the corresponding pre-set angle in advance, a set of equipment can complete the scanning work of 6 lanes.

[0132] According to an embodiment of the present invention, as Figure 3 shown, the container truck parking alignment system 200 at the terminal further includes: a display screen 207, and the display screen 207 is connected to the main control module 204; the display screen 207 is used to display the distance between the outer bounding box of the container truck sent by the main control module 204 and the target alignment plane.

[0133] It is understandable that the main control module 204 guides the container truck to align through the LED display screen 207, which is the main part of the software operation. After analysis and calculation by the software program of the industrial control computer 205, the main control module 204 will command the container truck through this LED display screen 207, including the precision range distance from the alignment position, whether the alignment is successful, waiting for the spreader to load and unload, whether it can safely drive away, etc. The LED display screen 207 is the key to the interaction between the control room (host computer 203) and the container truck driver. After calculating and analyzing key data, it transmits the operation action opinions to the container truck driver to ensure the safety of the driver and improve the operation efficiency.

[0134] Specifically, the start process includes software startup and ground parameter extraction. Then, the industrial control computer 205 waits for relevant information transmitted by the main control module 204, which is the preparation stage of the program. When the control room (host computer 203) transmits lane information through the main control module 204, the pan-tilt 206 starts to rotate, and the program enters the main loop to start collecting and processing data. Through the analysis and processing of these data, the industrial control computer 205 will calculate the real-time position of the container truck, and then display on the LED display screen 207 how far the current position is from the alignment position in real time, with the precision unit being centimeters. When the container truck stops at a position within the precision range, the LED display screen 207 will prompt the driver that it can stop at the current position. Similarly, through the analysis and processing of the scanned data and combined with the information of the main control module 204, the program will obtain the operation status of the spreader and the container to ensure the safety of the container truck and the driver. When the quay crane is working, it is necessary to ensure that the spreader is lifted to a certain height before allowing the container truck to leave under the quay crane. The traditional working method mainly relies on the container truck driver to visually estimate the height of the spreader by his own experience. Due to the error in observing through the rearview mirror, the driver may misjudge, resulting in the container truck driving away before the spreader reaches the safe height, thus damaging the lock head of the quay crane spreader. Therefore, the program also prevents the spreader from being dragged by the container truck.

[0135] Based on this, the combination of the lidar 201 and the camera 202 ensures the safety of the operation through double insurance. At the same time, the point cloud model will become more intuitive because of the auxiliary coloring of the camera 202, and the staff in the control room can directly and quickly master the on-site situation through this camera. In addition, with the addition of the camera, there is more room for subsequent software upgrades.

[0136] At the same time, the scanning device of the present invention includes a 2D lidar, a camera, and a pan-tilt with a programmable rotation angle, which can cover at least 6 lanes of the operation site, saving deployment and maintenance costs. Among them, the lidar can be a single-line lidar.

[0137] In addition to installing scanning devices under the quay crane, lidar or cameras can also be installed in front of and behind the foundation beams on the sea side and land side of the quay crane. This solution can also achieve the same effect and obtain the buckle information to determine whether the buckle connecting the shelf and the container is unlocked or not, preventing the staff from failing to adjust the buckle to the correct state due to negligence.

[0138] In summary, according to the method, device and system for parking and aligning terminal container trucks proposed in the embodiments of the present invention, the method includes obtaining point cloud data of the site where the container truck is to park; obtaining parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data; obtaining model data of the container truck according to the point cloud data, and obtaining contour data of the outer bounding box of the container truck according to the model data; obtaining the container type of the container truck according to the model data of the container truck; determining parameters of the target alignment plane where the head or tail of the container truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; obtaining a first distance between the outer bounding box of the container truck and the target alignment plane, and when the first distance is less than or equal to a preset distance, determining that the container truck has successfully parked and aligned, so as to realize the automation, digitization and intelligence of the parking and alignment of terminal container trucks.

[0139] Note that the above is only a preferred embodiment of the present invention and the technical principles applied. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described here. Various obvious changes, re-adjustments and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the concept of the present invention, more other equivalent embodiments can be included, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for parking and aligning terminal container trucks, characterized in that, Including the following steps: Obtain the point cloud data of the site where the container truck is to park; Obtain the parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data; Obtain the model data of the container truck according to the point cloud data, and obtain the contour data of the outer bounding box of the container truck according to the model data; Obtain the container type of the container truck according to the model data of the container truck; Determine the parameters of the target alignment plane where the head or tail of the container truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane; Obtain the first distance between the outer bounding box of the container truck and the target alignment plane, and when the first distance is less than or equal to a preset distance, determine that the container truck has successfully parked and aligned.

2. The method for parking and aligning terminal container trucks according to claim 1, characterized in that, The obtaining the parameters of the ground plane of the target parking space where the container truck is to park according to the point cloud data includes: Perform filtering processing on the point cloud data to obtain filtered point cloud data; Obtain the parameters of the ground plane of the target parking space according to the filtered point cloud data.

3. The method for parking and aligning terminal container trucks according to claim 2, characterized in that, The obtaining the parameters of the ground plane of the target parking space according to the filtered point cloud data includes: Define a segmentation model, and determine whether the current point in the filtered point cloud data is suitable for the segmentation model. If so, classify the current point as an inlier; if not, classify the current point as an outlier; Traverse the points in the filtered point cloud data, count the number of all inliers. When the number of inliers is greater than or equal to a preset threshold, update the segmentation model according to all the current inliers, and store all the inliers as the segmentation result; When the number of all inliers is less than the preset threshold, determine whether the number of all current inliers is greater than the number of inliers in the segmentation model. If so, update the segmentation model according to all the current inliers, and store all the inliers as the segmentation result; if not, use the segmentation model as the segmentation result; wherein, the segmentation model is a ground plane model; Obtain the parameters of the ground plane of the target parking space according to the segmentation result.

4. The method for parking and aligning terminal container trucks according to claim 1, characterized in that, The obtaining the model data of the container truck according to the point cloud data, and obtaining the contour data of the outer bounding box of the container truck according to the model data includes: Perform straight-through filtering processing on the point cloud data to obtain straight-through filtered point cloud data; Obtain the model data of the container truck according to the straight-through filtered point cloud data; Extract the contour data of the outer bounding box of the container truck according to the model data.

5. The method for parking and aligning terminal container trucks according to any one of claims 1-4, characterized in that, After determining that the container truck has successfully parked and aligned, it further includes: Model the container according to the point cloud data to obtain the model data of the container; Judge whether the container and the container truck are successfully aligned according to the model data of the container and the model data of the container truck.

6. The method for parking and aligning terminal container trucks according to claim 5, characterized in that, After the container and the container truck are successfully aligned, it further includes: Model the spreader through the point cloud data to obtain the model data of the spreader; Based on the model data of the spreader and the model data of the container, determine whether the spreader is separated from the container completely.

7. The method for parking and aligning terminal container trucks according to claim 1, 2, 3, 4 or 6, characterized in that, After obtaining the point cloud data of the site where the truck is to park, it further includes: Perform coloring processing on the point cloud data.

8. A system for parking and aligning terminal container trucks, characterized in that, For executing the quay truck parking alignment method according to any one of claims 1-7, including: A point cloud data acquisition module, configured to acquire the point cloud data of the site where the truck is to park; A ground plane parameter acquisition module, configured to acquire the parameters of the ground plane of the target parking space where the truck is to park according to the point cloud data; A truck model data acquisition module, configured to acquire the model data of the truck according to the point cloud data, and acquire the contour data of the outer bounding box of the truck according to the model data; A container type acquisition module, configured to acquire the container type of the truck according to the model data of the truck; A target alignment plane determination module, configured to determine the parameters of the target alignment plane to which the head or tail of the truck is to be aligned according to the container type and the parameters of the ground plane of the target parking space; the target alignment plane is perpendicular to the ground plane; A judgment module, configured to acquire the distance between the outer bounding box of the truck and the target alignment plane, and when the distance is less than or equal to a preset distance, judge that the truck parking alignment is successful.

9. A truck parking and alignment device for a wharf, characterized in that, It includes: A lidar, a camera, a host computer, a main control module, an industrial control computer and a pan-tilt head; The host computer is configured to send instructions to the main control module, the main control module forwards the instructions to the industrial control computer, and the industrial control computer adjusts the attitude angle of the pan-tilt head according to the instructions; the lidar and the camera are arranged on the pan-tilt head and are configured to scan the point cloud data of the target parking space where the truck is to park; the lidar and the camera are further configured to upload the acquired point cloud data to the industrial control computer, and the industrial control computer forwards the point cloud data to the main control module; The main control module executes the quay truck parking alignment method according to any one of claims 1-7 according to the point cloud data, so that the truck parks in the target parking space.

10. The truck parking and alignment device for a wharf according to claim 9, characterized in that, It further includes: A display screen, the display screen is connected to the main control module; the display screen is configured to display the distance between the outer bounding box of the truck sent by the main control module and the target alignment plane.

Images