Container positioning method, system and device based on laser radar and storage medium

Abstract

The invention provides a container positioning method, system and device based on a laser radar and a storage medium, and the method comprises the steps: obtaining first point cloud data covering a container and a loading and unloading platform, carrying out the iterative fitting of the first point cloud data through an RANSAC algorithm, obtaining plane equations corresponding to a plurality of dominant planes of the segmented container, and carrying out the calculation of the plane equations, calculating a current pose of the container in the reference coordinate system according to a plane equation of the plurality of dominant planes, and determining a first pose parameter; in the moving process of the container or the loading and unloading platform, second point cloud data obtained by scanning the tail environment of the container are obtained, local feature extraction and clustering filtering are conducted on the second point cloud data through an eight-neighborhood image algorithm to determine the tail position of the container, the first pose parameter is corrected according to the tail position of the container, and the tail position of the container is obtained. And outputting the second pose parameter of the container. According to the invention, multiple laser radars are matched, the real-time performance of scanning detection and the integrity of point cloud are ensured, and the precision and efficiency of container pose measurement are improved.

Description

Technical Field

[0001] The present invention relates to the field of radio detection technology, and in particular to a container positioning method, system, device and storage medium based on laser radar. Background Art

[0002] Current container loading and unloading positioning technology primarily relies on manual vision or single-lidar scanning, which presents significant limitations. Due to the limited vertical field of view, a single lidar cannot fully capture the top and bottom contours of a container, resulting in missing point cloud data and affecting plane fitting accuracy. Furthermore, the low scanning frequency of a single lidar generates a sparse point cloud, making it difficult to meet the requirements of high-precision pose calculations, especially in long-distance or large-scale scenarios.

[0003] During container loading and unloading, the relative position of the loading platform and container still requires manual calibration of reference points, which introduces human error and is unable to adapt to dynamic adjustments. Furthermore, the low scanning frequency of a single radar and the mechanical rotation required cause delays in vehicle position updates, making real-time tracking impossible. These issues collectively limit the accuracy and efficiency of automated loading and unloading, necessitating the use of technologies such as multi-sensor fusion, automatic calibration, and real-time processing to overcome these bottlenecks. Summary of the Invention

[0004] The embodiments of the present invention provide a lidar-based container positioning method, system, device, and storage medium to address the problems existing in related technologies. The technical solutions are as follows: In a first aspect, an embodiment of the present invention provides a container positioning method based on laser radar, comprising: Acquire first point cloud data covering the container and the loading and unloading platform, extract the platform point set of the loading and unloading platform in the first point cloud data, perform plane ground fitting, and establish a reference coordinate system based on the platform ground; Iteratively fit the container point cloud in the first point cloud data using the RANSAC algorithm to obtain the plane equations corresponding to multiple dominant planes of the segmented container. Calculate the current pose of the container in the reference coordinate system based on the plane equations of the multiple dominant planes to determine the first pose parameter. During the movement of the container or loading and unloading platform, the second point cloud data obtained by scanning the environment at the rear of the container is obtained. Local features of the second point cloud data are extracted using the eight-neighborhood image algorithm to obtain a feature set. The feature set is clustered and filtered to determine the rear position of the container. The first pose parameters are corrected according to the rear position of the container, and the second pose parameters of the container are output.

[0005] In one embodiment, the first point cloud data is obtained by scanning a container parked at a specified position using a single-line laser radar equipped with a rotating pan-tilt platform; wherein the rotating pan-tilt platform drives the single-line laser radar to rotate and scan at a constant angle.

[0006] In one embodiment, iteratively fitting the container point cloud in the first point cloud data using the RANSAC algorithm to obtain plane equations corresponding to multiple dominant planes of the segmented container includes: The container point cloud is preprocessed based on the voxel grid downsampling algorithm to obtain the preprocessed container point cloud; The preprocessed container point cloud is iteratively fitted using the RANSAC algorithm and preset plane fitting constraints to obtain the plane equations corresponding to multiple dominant planes. Among them, the plane fitting constraints include the deviation of the angle between adjacent planes being less than a preset angle and the plane roughness being less than a preset value.

[0007] In one embodiment, calculating the current posture of the container in the reference coordinate system based on the plane equations of the plurality of dominant planes, and determining the first posture parameter includes: Calculate container dimensions based on plane equations of multiple dominant planes, where the container dimensions include container length, width, and height; Solve the intersecting plane equations to obtain the coordinates of all vertices of the container, and determine the coordinates of the center of the container based on the coordinates of all vertices; The tilt angle of the container relative to the ground is calculated by the dot product of the bottom normal vector and the ground normal vector, and the horizontal yaw angle and pitch angle are determined according to the slopes of the plane equations of multiple dominant planes.

[0008] In one embodiment, it further includes: Based on the plane equations corresponding to multiple dominant planes, a corresponding spatial hierarchical index structure is constructed, and the BFS algorithm is used to traverse the spatial hierarchical index structure layer by layer; The neighborhood points are clustered according to the set cluster radius threshold to obtain multiple cluster point cloud groups; The size threshold is used to identify the interference objects in each cluster point cloud group, determine the location of the interference objects and remind the interference objects.

[0009] In one embodiment, local features are extracted from the second point cloud data using an eight-neighborhood image algorithm, and the obtained feature set includes: Extracting a point cloud of the container tail area from the second point cloud data, and projecting the tail point cloud onto a two-dimensional plane to generate a grid matrix; Based on the network matrix, the point cloud features of the eight neighborhood directions of the central grid are checked to obtain a feature set.

[0010] In one embodiment, it further includes: The point clouds in the front and back, up and down directions of the container in the second point cloud data are filtered, and the point clouds on the left and right sides of the container are retained. The two side plane equations are fitted using the minimum multiplication method, and the real-time direction of the container is determined based on the two side plane equations; among them, the two side plane equations are combined with the tail position of the container to determine the three-dimensional spatial coordinates of the entire container.

[0011] In a second aspect, an embodiment of the present invention provides a container positioning system based on laser radar, comprising: A single-line laser radar equipped with a rotating pan-tilt head is used to scan containers parked at designated locations and output the first point cloud data. Multi-line laser radar, used to scan the environment behind the container when the container or loading platform moves, and output second point cloud data; The server communicates with the single-line laser radar and the multi-line laser radar signals, and is used to execute the above-mentioned laser radar-based container positioning method.

[0012] In a third aspect, embodiments of the present invention provide an electronic device comprising: a memory and a processor. The memory and the processor communicate with each other via an internal connection path, the memory is configured to store instructions, and the processor is configured to execute the instructions stored in the memory. When the processor executes the instructions stored in the memory, the processor performs the method according to any of the aforementioned embodiments.

[0013] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a computer, the method in any one of the above-mentioned embodiments is executed.

[0014] The advantages or beneficial effects of the above technical solution include at least: The present invention uses a rotating single-line laser radar to scan the container and loading platform without blind spots to obtain first point cloud data. The first point cloud data is then plane-fitted using the RANSAC algorithm to determine the container's first pose parameters. When the container or loading platform moves during loading and unloading, a multi-line laser radar is used to scan the container's rear area in real time to obtain second point cloud data. The second point cloud data accurately locates the container's rear position, and the first pose parameters are corrected based on the real-time container rear position, outputting the container's accurate position during movement. The present invention combines the single-line laser radar with the multi-line laser radar to ensure real-time scanning and detection, as well as point cloud integrity, improving the accuracy and efficiency of container pose measurement.

[0015] The above summary is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments and features described above, further aspects, embodiments and features of the present invention will be readily apparent by reference to the accompanying drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the multiple drawings represent the same or similar components or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments disclosed herein and should not be construed as limiting the scope of the invention.

[0017] Figure 1 Schematic diagram of the process of the container positioning method based on laser radar of the present invention; Figure 2 This is a schematic diagram of the positional relationship between the loading and unloading platform and the container of the present invention; Figure 3 This is a schematic diagram of a three-dimensional model of a container according to the present invention; Figure 4 This is a schematic diagram of positioning interference objects in a container according to the present invention; Figure 5 FIG. 1 is a structural block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] Hereinafter, only certain exemplary embodiments are briefly described. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and description are to be considered as illustrative in nature and not restrictive.

[0019] Example 1 This embodiment provides a lidar-based container positioning method, which relates to the field of intelligent warehousing and logistics technology. Through dynamic scanning and intelligent point cloud processing, it realizes autonomous coordinate construction of the loading plane, full-size measurement of the container, real-time positioning of the vehicle posture and interference warning. It is suitable for intelligent container loading and unloading scenarios in automated ports, railway freight stations and logistics centers.

[0020] It should be explained that in this embodiment, the container is transported by a vehicle. After the container is fixed on the vehicle, it is transported to a designated location around the loading and unloading platform. A laser radar is used to scan the container on the vehicle and the loading and unloading platform, and the position of the container is located based on the scanning results.

[0021] To further improve positioning accuracy, this embodiment uses a retractable bracket to install a rotating pan-tilt platform on the ceiling directly opposite the designated location. A single-line LiDAR is mounted on the rotating pan-tilt platform. This high-refresh-rate (≥50Hz) single-line LiDAR is equipped with a 360° continuously rotating pan-tilt platform with a scanning radius of ≥20m, covering the entire surface of the container and the loading and unloading platform, thereby generating a dense point cloud (point cloud density >200 points / m²). Furthermore, a multi-line LiDAR with a field of view of 120°*50° is fixedly mounted on the loading and unloading platform. This multi-line LiDAR dynamically scans the moving container, enabling real-time positioning of the container and subsequent intelligent container loading and unloading operations.

[0022] It should be noted that a single-line laser radar refers to a laser radar device that only scans one laser beam in the vertical direction. Compared with a single-line laser radar, a multi-line laser radar emits multiple laser beams at the same time. These laser beams are distributed at different vertical angles, thereby being able to simultaneously obtain multiple levels of distance information to form three-dimensional point cloud data.

[0023] Under the above radar deployment, if Figure 1 As shown, the specific steps of the container positioning method based on laser radar in this embodiment include: Step S1: Acquire first point cloud data covering the container and the loading and unloading platform, extract the platform point set of the loading and unloading platform in the first point cloud data, perform plane ground fitting, and establish a reference coordinate system based on the platform ground.

[0024] In this embodiment, when the vehicle with the container installed stops at a designated location, the rotating pan-tilt table drives the single-line laser radar to rotate and scan at a constant angular velocity, generating the first point cloud data covering the container and the loading and unloading platform. The first point cloud data contains the timestamp, position and posture information of each point cloud data. The precise position and posture corresponding to the point cloud sampling time (R t , T t ), where R t represents the rotation matrix, T t Represents the translation vector, and then uses the pose at the reference moment (R ref , T ref ), convert the point cloud data at the sampling moment into the reference coordinate system, and the calculation formula is as follows: p ref =R ref -1×(R t ×p t +T t -T ref );in: p t Represents the point cloud coordinates at the sampling moment; R t , Tt are the rotation matrix and translation vector at the sampling moment; R ref , T ref is the rotation matrix and translation vector at the reference moment; p ref It is the point cloud coordinates in the reference coordinate system after motion compensation.

[0025] In this embodiment, all motion-compensated point cloud data are aligned to the same reference coordinate system, and the ICP algorithm or other registration methods are used for further fine registration, thereby fusing multiple frames of scan data into a dense point cloud dataset covering the entire surface of the container and the ground. The dense point cloud dataset is used to construct a complete 3D model of the container (such as Figure 3 As shown), the three-dimensional imaging display of the container is completed.

[0026] Then, the platform point set P of the loading and unloading platform is extracted from the first point cloud data / dense point cloud data set according to the preset tolerance threshold. ground ={p i ∣z i ∈[h- , h+ ]}(h is the platform height, The platform point set is fitted using the least squares method to obtain the platform ground equation ax+by+cz+d=0, and the platform reference coordinate system is constructed based on the platform ground equation. In this embodiment, the platform point set is extracted based on the tolerance threshold to establish the reference coordinate system, which can eliminate the coordinate drift caused by the vibration of the loading and unloading platform to a certain extent, and the posture measurement stabilization time is less than 0.5 seconds. Figure 2 As shown, in this embodiment, the midpoint of the short side of the loading and unloading platform is used as the origin to establish a reference coordinate system.

[0027] Step S2: Iteratively fit the container point cloud in the first point cloud data through the RANSAC algorithm to obtain the plane equations corresponding to multiple dominant planes of the segmented container, calculate the current position of the container in the reference coordinate system based on the plane equations of the multiple dominant planes, and determine the first pose parameter.

[0028] In order to improve computational efficiency, the first point cloud data is preprocessed based on the voxel grid downsampling algorithm, that is, the three-dimensional space is divided into small cubic grids of equal size (such as 5cm×5cm×5cm). Each grid is called a voxel. In each non-empty voxel grid, all the original points falling within the grid are collected, and the geometric center (or centroid) of these points is used to represent the points of the entire grid. Finally, only one representative point is retained for each voxel.

[0029] In this embodiment, the container point cloud in the first point cloud data is preprocessed based on the voxel grid downsampling algorithm to obtain a preprocessed container point cloud. The RANSAC (Random Sample Consensus) algorithm is used to iteratively fit multiple dominant planes in the preprocessed container point cloud to obtain plane equations corresponding to the multiple dominant planes of the segmented container. The current pose of the container in the reference coordinate system is then calculated based on the plane equations to determine the first pose parameter.

[0030] It should be noted that the first posture parameter includes container size, container vertex coordinates, center coordinates, relative inclination angle, horizontal yaw angle, and pitch angle.

[0031] Specifically, this example uses the RANSAC algorithm to iteratively fit the five dominant planes in the container point cloud. The five dominant planes are the front, left, right, top, and bottom planes of the container. The plane fitting constraints are set to the angle deviation between adjacent planes < 3° and the plane roughness < 0.005. The plane equations corresponding to the five dominant planes are: S front : A1x+B1y+C1z=D1; ... S bottom : A5x+B5y+C5z=D5.

[0032] On the basis of determining the plane equations of the five dominant planes, the vertex coordinates of the container are determined by solving the intersection of the three intersecting planes. The vertex coordinates V k The calculation formula is: ; Where m is the index of the intersecting three planes.

[0033] For example, for vertex V1, suppose it is formed by the front plane S front , left plane S left and upper plane S top The intersection results in: ; By solving this linear equation system, we can obtain the coordinates (x1, y1, z1) of vertex V1. Similarly, we can find the coordinates of the other seven vertices, and thus determine the coordinates of the container center based on the coordinates of all the vertices.

[0034] After determining the vertex coordinates of the container, the length of the container can be indirectly calculated based on the container size specifications or the distance between other planes (such as sides or vertices). A similar method can also be used to calculate the width and height of the container to obtain the container size.

[0035] In addition, the relative tilt angle of the container can be obtained by calculating the angle between the normal vector of the container and the normal vector of the ground. The tilt angle θ of the container relative to the ground is: θ=arccos(Ncontainer•Nground); where, Ncontainer: the normal vector of a dominant plane of the container (such as the bottom); Nground: The normal vector of the ground plane (usually (0,0,1)(0,0,1)); The dot product result reflects the cosine of the angle between the two normal vectors, and arccos obtains the actual inclination angle (radians or degrees).

[0036] In addition, the horizontal yaw angle and the pitch angle can also be determined based on the plane equations of multiple dominant planes, where the slope of the container side equation (i.e., the plane equation of the left plane or the plane equation of the right plane) is the horizontal yaw angle, and the slope of the upper and lower surface equations (i.e., the plane equation of the upper plane or the plane equation of the lower plane) is the pitch angle.

[0037] Furthermore, it is necessary to determine whether there are any interference objects (unexpected objects) in the container in order to facilitate integration into the automated loading and unloading system. In this embodiment, interference object detection is required during the container positioning process. The specific method is as follows: Based on the plane equations corresponding to multiple dominant planes, the corresponding spatial hierarchical index structure KD-TREE is constructed.

[0038] It should be noted that KD-Tree is a k-dimensional binary search tree that organizes point cloud data into a hierarchical structure by recursively dividing the space along the coordinate axis to support efficient radius neighborhood search.

[0039] The breadth-first search BFS algorithm is used to traverse the spatial hierarchical index structure layer by layer. The BFS traversal needs to dynamically call the neighborhood query function of the KD-Tree and cluster the neighborhood points according to the set clustering radius threshold. That is, each time a point p is taken from the queue, its r is quickly obtained through the KD-Tree. cluster All points in the neighborhood are obtained, thus obtaining multiple cluster point cloud groups. cluster It needs to be slightly larger than the average spacing of the point cloud to cover the continuity of the object surface and avoid mistakenly merging independent objects.

[0040] Use the size threshold to identify the interference objects in each clustered point cloud group, determine the location of the interference objects and remind the interference objects. Specifically, for each clustered point cloud group, calculate the size of its axial bounding box: Length L = max(x) - min(x) Width W = max(y) - min(y) Height H = max(z) - min(z) If L, W, and H are all less than 30 cm, the clustered point cloud group is classified as an interference object (such as tools, debris, etc.), triggering cleaning or alarming; otherwise, it is retained as a valid object (such as a cargo box, pallet, etc.). Figure 4 As shown, Figure 4 The red box in the middle shows the location of the interference object in the container.

[0041] Step S3: During the movement of the container or loading and unloading platform, obtain the second point cloud data obtained by scanning the environment at the rear of the container, perform local feature extraction on the second point cloud data through the eight-neighborhood image algorithm to obtain a feature set, perform clustering and filtering on the feature set to determine the rear position of the container, correct the first pose parameters according to the rear position of the container, and output the second pose parameters of the container.

[0042] In order to achieve the purpose of loading and unloading, the position of the container or loading and unloading platform needs to be moved and adjusted so that the container and the loading and unloading platform are in the right position, so as to complete the subsequent loading and unloading operations. During the movement of the container or the loading and unloading platform, the rear position of the vehicle is scanned by the multi-line laser radar installed on the loading and unloading platform, thereby obtaining a multi-line laser radar real-time frame 3D point cloud set, also known as the second point cloud data. Figure 2 As shown, Figure 2 The container is mainly aligned with the loading and unloading platform by moving the loading and unloading platform to facilitate subsequent cargo loading and unloading.

[0043] The car body point clouds of the front, rear, upper, and lower planes are filtered from the second point cloud data, and the point cloud data of the two side planes of the vehicle are retained. The two side planes of the vehicle are fitted using the least squares method to obtain the plane equations of the two side planes (i.e., the left plane and the right plane). The rotation posture of the car body is determined according to the slopes of the plane equations of the two side planes, thereby determining the real-time direction of the container.

[0044] Since the multi-line laser radar mainly scans the rear position of the vehicle, the number of point clouds in the rear area of ​​the container in the second point cloud data is relatively concentrated. At this time, the point cloud of the rear area of ​​the container in the second point cloud data is extracted, and the point cloud of the tail area is projected onto a two-dimensional plane. The point cloud of the tail area is divided into grids according to a certain resolution and converted into a two-dimensional matrix form to obtain a grid matrix; each grid cell contains the point cloud information falling within the area, which can be characteristic values ​​such as the number of points and average height.

[0045] During the container tail location determination process, this embodiment uses an eight-neighborhood algorithm to perform feature search and analysis on gridded point cloud data. By calculating and comparing the feature values ​​of each grid cell and its eight-neighborhood, regions with specific geometric or density characteristics, such as edges and corners, are identified, resulting in several feature sets. For each pixel (or grid cell), its eight-neighborhood includes the pixel itself and its eight surrounding pixels (upper, lower, left, right, upper left, upper right, lower left, and lower right).

[0046] By analyzing the feature distribution within the eight-neighborhood, outliers or noise points can be detected and removed, thereby improving the accuracy and robustness of feature extraction. The details are as follows: Statistics are collected for the number of points or density in each grid cell and its eight neighborhoods. If the density of a grid cell is significantly lower than the average value of its neighborhood, it may be a noise point or an outlier and can be considered for removal.

[0047] Calculate the height difference (i.e., z-axis coordinate) of each grid cell compared to the heights of its eight neighboring cells. If the height change of a grid cell is much greater than the average height change of its neighboring cells, this may be due to an isolated outlier or noise and should be removed.

[0048] We can also analyze the connectivity between grid cells to identify isolated points that are not connected to any other cells. These isolated points are likely to be noise points and should be removed.

[0049] In the data after interference points are removed, pattern matching is performed on the remaining point cloud data based on the unique geometric shape of the container tail (such as rectangle, trapezoid, etc.), so as to accurately determine the exact position of the container tail.

[0050] This embodiment uses a multi-line laser radar to scan the vehicle in real time while it is moving. The real-time orientation of the container is determined by fitting the two side planes of the vehicle. The three-dimensional coordinates of the container's rear are determined using an eight-neighborhood algorithm, thereby accurately positioning the container's real-time posture. The first posture parameters of the container in step S2 are corrected based on the container's real-time posture to obtain the container's second posture parameters in the current state. These second posture parameters include the container's center coordinates, relative inclination, horizontal yaw angle, and pitch angle. Simultaneously, the position of interfering objects within the container is accurately located based on the container's real-time posture. The container's center coordinates, dimensional data, horizontal yaw angle, pitch angle, and the coordinates of the interfering objects within the container are output to the automatic loading and unloading platform control system, enabling precise automatic loading and unloading of cargo. For example, the robot arm is automatically controlled to rotate and move to a specified position, forking cargo for automatic loading, thereby improving loading and unloading efficiency. In actual unloading scenarios, in addition to using multi-line lidar to scan and locate containers, it is also possible to scan pallets inside the container, separate the bottom surface of the pallet and the container wall through the RANSAC algorithm, segment individual pallets based on point cloud density, and extract edges from the pallet surface to locate the coordinates of the pallet holes, which are used to guide the fork arm for precise unloading.

[0051] The beneficial effects of this embodiment are as follows: 1. Full-view, zero-delay coverage: This embodiment utilizes a gimbal to drive a single-line LiDAR to achieve 360° rotation, enabling scanning without blind spots. Furthermore, multi-line LiDARs perform real-time scanning while the vehicle is moving to accurately locate the container's real-time position, ensuring both real-time performance and point cloud integrity. 2. Complex Scene Adaptation: The gimbal drives a single-line LiDAR. Compared to the traditional method of using a single-line LiDAR to intercept a single cross-section to calculate the slope and attitude, this LiDAR fits the entire container plane. This eliminates the impact of local container deformation, avoids misjudgment of the overall attitude, and adapts to complex scenarios. 3. High-precision modeling: When fitting the container plane using RANSAC, even if there are depressions or attachments (outliers) on the surface, the overall plane pose can still be accurately estimated through inlier consensus. The RANSAC plane fitting error is less than 2mm, and the dimensional measurement accuracy reaches the centimeter level (measured error is less than 1cm in 99.5% of cases). 4. Dynamic benchmark construction: The platform point set of the loading and unloading platform is extracted through the tolerance threshold to establish a benchmark coordinate system, which can eliminate the coordinate drift caused by the vibration of the loading and unloading platform. The pose measurement stabilization time is less than 0.5 seconds. 5. Intelligent anti-interference: KD-Tree combined with BFS clustering algorithm is used to identify interference objects in containers, with an accuracy rate of >95% and a false alarm rate of <3%.

[0052] Example 2 This embodiment provides a container positioning system based on laser radar, which includes: A single-line LiDAR equipped with a rotating gimbal is used to scan containers parked at designated locations. The gimbal drives the single-line LiDAR to rotate and scan at a constant angular velocity, outputting the first point cloud data covering the entire surface of the container and the loading and unloading platform. Multi-line laser radar, installed on the loading and unloading platform, is used to scan the environment behind the container when the container or loading and unloading platform moves, and output the second point cloud data; The server communicates signals with the single-line laser radar and the multi-line laser radar, and is used to execute the laser radar-based container positioning method as described in Example 1.

[0053] It should be noted that the various functions of the system in the embodiment of the present invention can be found in the corresponding description of the above method and will not be repeated here.

[0054] Example 3 This embodiment provides an electronic device, Figure 5 FIG. 1 shows a structural block diagram of an electronic device according to an embodiment of the present invention. Figure 5 As shown, the electronic device includes a memory 100 and a processor 200. The memory 100 stores a computer program that can be executed on the processor 200. When the processor 200 executes the computer program, the lidar-based container positioning method described in the above embodiment is implemented. The number of the memory 100 and the processor 200 can be one or more.

[0055] The electronic device also includes: The communication interface 300 is used to communicate with external devices and perform data exchange transmission.

[0056] If the memory 100, processor 200, and communication interface 300 are implemented independently, they can be connected to each other via a bus and communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. This bus can be divided into an address bus, a data bus, a control bus, etc.

[0057] Optionally, in a specific implementation, if the memory 100, the processor 200 and the communication interface 300 are integrated on a chip, the memory 100, the processor 200 and the communication interface 300 can communicate with each other through an internal interface.

[0058] An embodiment of the present invention provides a computer-readable storage medium storing a computer program. When the program is executed by a processor, the method provided in the embodiment of the present invention is implemented.

[0059] An embodiment of the present invention further provides a chip, which includes a processor for calling and executing instructions stored in a memory, so that a communication device equipped with the chip executes the method provided by the embodiment of the present invention.

[0060] An embodiment of the present invention also provides a chip, comprising: an input interface, an output interface, a processor and a memory, wherein the input interface, the output interface, the processor and the memory are connected via an internal connection path, and the processor is used to execute the code in the memory. When the code is executed, the processor is used to execute the method provided by the embodiment of the invention.

[0061] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. It is worth noting that the processor may be a processor that supports the advanced reduced instruction set machine (ARM) architecture.

[0062] Furthermore, optionally, the above-mentioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Among them, the non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM) or flash memory. The volatile memory may include random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0063] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the present invention are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0064] In the description of this specification, reference to the terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless otherwise inconsistent.

[0065] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0066] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various modifications and substitutions within the technical scope disclosed in the present invention, and such modifications and substitutions are intended to be encompassed by the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A container positioning method based on laser radar, characterized in that: include: Acquire first point cloud data covering the container and the loading and unloading platform, extract the platform point set of the loading and unloading platform from the first point cloud data, perform plane ground fitting, and establish a reference coordinate system based on the platform ground; Iteratively fitting the container point cloud in the first point cloud data using the RANSAC algorithm to obtain plane equations corresponding to multiple dominant planes of the segmented container, calculating the current pose of the container in the reference coordinate system based on the plane equations of the multiple dominant planes, and determining the first pose parameter; During the movement of the container or loading and unloading platform, second point cloud data obtained by scanning the rear environment of the container is obtained, local features of the second point cloud data are extracted using an eight-neighborhood image algorithm to obtain a feature set, clustering and filtering are performed on the feature set to determine the rear position of the container, the first pose parameters are corrected according to the rear position of the container, and the second pose parameters of the container are output.

2. The container positioning method based on laser radar according to claim 1, characterized in that: The first point cloud data is obtained by scanning a container parked at a specified position using a single-line laser radar equipped with a rotating pan-tilt platform, wherein the rotating pan-tilt platform drives the single-line laser radar to rotate and scan at a constant angle.

3. The container positioning method based on laser radar according to claim 1, characterized in that: The iterative fitting of the container point cloud in the first point cloud data by the RANSAC algorithm to obtain the plane equations corresponding to the multiple dominant planes of the segmented container includes: The container point cloud is preprocessed based on the voxel grid downsampling algorithm to obtain the preprocessed container point cloud; The preprocessed container point cloud is iteratively fitted using the RANSAC algorithm and preset plane fitting constraints to obtain the plane equations corresponding to multiple dominant planes. Among them, the plane fitting constraints include the deviation of the angle between adjacent planes being less than a preset angle and the plane roughness being less than a preset value.

4. The container positioning method based on laser radar according to claim 1, characterized in that: Calculating the current posture of the container in the reference coordinate system according to the plane equations of the plurality of dominant planes and determining the first posture parameter includes: Calculate container dimensions based on plane equations of multiple dominant planes, where the container dimensions include container length, width, and height; Solve the intersecting plane equations to obtain the coordinates of all vertices of the container, and determine the coordinates of the center of the container based on the coordinates of all vertices; The tilt angle of the container relative to the ground is calculated by the dot product of the bottom normal vector and the ground normal vector, and the horizontal yaw angle and pitch angle are determined according to the slopes of the plane equations of multiple dominant planes.

5. The container positioning method based on laser radar according to claim 1, characterized in that: Also includes: Based on the plane equations corresponding to multiple dominant planes, a corresponding spatial hierarchical index structure is constructed, and the BFS algorithm is used to traverse the spatial hierarchical index structure layer by layer; The neighborhood points are clustered according to the set cluster radius threshold to obtain multiple cluster point cloud groups; The size threshold is used to identify the interference objects in each cluster point cloud group, determine the location of the interference objects and remind the interference objects.

6. The container positioning method based on laser radar according to claim 1, characterized in that: The local feature extraction of the second point cloud data by the eight-neighborhood image algorithm is performed to obtain a feature set including: Extracting a point cloud of the container tail area from the second point cloud data, and projecting the tail point cloud onto a two-dimensional plane to generate a grid matrix; Based on the network matrix, the point cloud features of the eight neighborhood directions of the central grid are checked to obtain a feature set.

7. The container positioning method based on laser radar according to claim 1, characterized in that: Also includes: The point clouds in the front and back, and up and down directions of the container in the second point cloud data are filtered, and the point clouds on the left and right sides of the container are retained. The two side plane equations are fitted using the minimum multiplication method, and the real-time direction of the container is determined based on the two side plane equations; wherein the two side plane equations are combined with the tail position of the container to determine the three-dimensional spatial coordinates of the entire container.

8. A container positioning system based on laser radar, characterized in that: include: A single-line laser radar equipped with a rotating pan-tilt head is used to scan containers parked at designated locations and output the first point cloud data. Multi-line laser radar, used to scan the environment behind the container when the container or loading platform moves, and output second point cloud data; The server communicates with the single-line laser radar and the multi-line laser radar signals, and is used to execute the laser radar-based container positioning method as described in any one of claims 1 to 7.

9. An electronic device, characterized in that: include: A processor and a memory, wherein the memory stores instructions, and the instructions are loaded and executed by the processor to implement the container positioning method based on laser radar as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the lidar-based container positioning method according to any one of claims 1 to 7 is implemented.

Images