Vehicle unloading method and related equipment

Abstract

The invention provides a vehicle unloading method and related equipment, and the vehicle unloading method comprises the steps: obtaining point cloud data obtained by scanning an environment where a vehicle is located through a laser radar in the vehicle and pose data of the vehicle under the condition that the vehicle has an unloading demand; determining the priority of each unloading area of the material storage yard where the vehicle is located; according to the point cloud data, the pose data, the priority and a prediction model in the vehicle, predicting a driving track of the vehicle in the material storage yard; and the vehicle is controlled to run to the material pile where the end point of the running track is located according to the running track for discharging. According to the invention, the unloading efficiency of the vehicle is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a vehicle unloading method and related equipment. Background Technology

[0002] In the industrial transportation sector, especially in special scenarios such as waste disposal and mining, vehicles are required to frequently perform precise unloading operations. Unloading operations typically require vehicles to park at specific locations next to the material pile, maintaining a precise distance and angle from the pile to ensure safety and efficiency during the unloading process.

[0003] In the exemplary technology, the driver determines the unloading location of the material pile, and then the driver drives the vehicle to the unloading location to unload the material.

[0004] However, drivers determine the unloading position of the material pile based on their own experience. Inexperienced drivers need to repeatedly drive the vehicle to adjust the position, resulting in a long unloading time and low vehicle efficiency. Summary of the Invention

[0005] Based on the above-mentioned technological status, this application provides a vehicle unloading method and related equipment to solve the problem of low vehicle efficiency.

[0006] To achieve the above-mentioned technical objectives, this application proposes the following technical solution: In a first aspect, this application provides a vehicle unloading method, including: When the vehicle has a need to unload, the point cloud data obtained by the lidar in the vehicle scanning the environment where the vehicle is located and the pose data of the vehicle are acquired. Determine the priority of each unloading area in the stockpile where the vehicle is located; Based on the point cloud data, the pose data, the priority, and the prediction model in the vehicle, the driving trajectory of the vehicle in the stockpile is predicted. Control the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading.

[0007] In some implementations, predicting the vehicle's trajectory in the stockpile based on the point cloud data, the pose data, the priority, and the prediction model in the vehicle includes: The point cloud data is corrected based on the pose data to obtain intermediate point cloud data, and the height of each material pile is extracted from the intermediate point cloud data. A height matrix is ​​constructed based on the height of each extracted material pile, and a priority matrix is ​​constructed based on each of the aforementioned priorities; The height matrix and the priority matrix are fused to obtain a fusion matrix, and the fusion matrix is ​​input into the prediction model to obtain the vehicle's driving trajectory in the stockpile.

[0008] In some implementations, constructing a height matrix based on the heights of each extracted stockpile includes: Based on the height of each extracted material pile, construct the matrix to be processed; The matrix to be processed is subjected to time-domain filtering and spatial-domain noise reduction in sequence to obtain the height matrix.

[0009] In some implementations, the step of correcting the point cloud data based on the pose data includes: Obstacle data of the vehicle's environment collected by the millimeter-wave radar in the vehicle are obtained, and the point cloud data is dynamically filtered for obstacles based on the obstacle data to obtain processed point cloud data, which includes point clouds corresponding to multiple material piles. The point cloud of the processed point cloud data is corrected based on the pose data.

[0010] In some embodiments, controlling the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading includes: Determine the driving direction of the vehicle at each path point on the driving trajectory and the target direction corresponding to each path point; Based on the driving direction and target direction corresponding to each path point, the driving trajectory is divided into multiple sub-trajectories. The directional angle between the target directions of adjacent path points on the sub-trajectories is less than a first directional angle threshold, and the directional angle between the driving directions of adjacent path points on the sub-trajectories is less than a second directional angle threshold. The vehicle is controlled to travel along each of the sub-trajectories to the material pile at the end of the travel trajectory for unloading, wherein the vehicle speed and gear remain unchanged while traveling on the sub-trajectories.

[0011] In some embodiments, controlling the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading includes: Obtain the kinematic constraints of the vehicle; Based on the kinematic constraints, the vehicle trajectory is processed to obtain the target trajectory; Control the vehicle to travel along the target trajectory to the material pile at the end of the target trajectory for unloading.

[0012] In some embodiments, before acquiring the point cloud data obtained by the lidar in the vehicle scanning the environment where the vehicle is located and the vehicle's pose data, the method further includes: Acquire various training samples, including a material pile map containing multiple material pile regions and the priority of each material pile region in the material pile map, which is drawn based on point cloud data collected by LiDAR; The preset model is trained based on each of the training samples to obtain the prediction model.

[0013] Secondly, this application provides a vehicle, comprising: The acquisition module is used to acquire point cloud data obtained by the lidar in the vehicle scanning the environment where the vehicle is located, as well as the vehicle's pose data, when the vehicle has a need to unload. The module determines the priority of each unloading area in the stockpile where the vehicle is located; The prediction module is used to predict the driving trajectory of the vehicle in the stockpile based on the point cloud data, the pose data, the priority, and the prediction model in the vehicle. The control module is used to control the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading.

[0014] Thirdly, this application provides an electronic device, including a memory and a processor, wherein, The memory is connected to the processor and is used to store programs; The processor is used to implement the vehicle unloading method as described in the first aspect or any implementation thereof by running a program in the memory.

[0015] Fourthly, this application provides a computer program product, which, when executed by a processor, implements the vehicle unloading method as described in the first aspect or any implementation thereof.

[0016] Fifthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the vehicle unloading method as described in the first aspect or any implementation thereof.

[0017] This application provides a vehicle unloading method and related equipment. When a vehicle needs to unload, it acquires point cloud data obtained from the vehicle's environment scanned by a LiDAR scanner, as well as the vehicle's pose data. The priority of each unloading area in the stockpile where the vehicle is located is determined. Based on the priority, point cloud data, pose data, and a prediction model, the vehicle's trajectory in the stockpile is predicted, and the vehicle is controlled to travel along the trajectory to the stockpile at the end of the trajectory for unloading. In this application, by using point cloud data from the vehicle's LiDAR scanned environment, the vehicle's pose data, the priority of each stockpile area in the stockpile, and a prediction model, the unloading trajectory is automatically and accurately planned for the vehicle. This allows the vehicle to accurately and quickly reach the stockpile for unloading based on the trajectory, improving the vehicle's unloading efficiency. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 A flowchart of a vehicle unloading method provided in this application embodiment Figure 1 .

[0020] Figure 2 A flowchart of a vehicle unloading method provided in this application embodiment Figure 2 .

[0021] Figure 3 A flowchart of a vehicle unloading method provided in this application embodiment Figure 3 .

[0022] Figure 4 A flowchart of a vehicle unloading method provided in this application embodiment Figure 4 .

[0023] Figure 5 A flowchart of a vehicle unloading method provided in this application embodiment Figure 5 .

[0024] Figure 6 This is a structural schematic diagram of a vehicle provided in an embodiment of this application.

[0025] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0026] The technical solution proposed in this application is applicable to vehicle unloading scenarios and aims to solve the problem of low unloading efficiency of vehicles. By employing the technical solution described in this application, point cloud data of the environment scanned by the LiDAR in the vehicle, the vehicle's pose data, the priority of each material pile area in the stockpile, and a prediction model, the vehicle's unloading trajectory is automatically and accurately planned. This allows the vehicle to accurately and quickly reach the material pile for unloading based on the trajectory, thereby improving the vehicle's unloading efficiency.

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] It should be noted that the user information (including but not limited to electrical equipment information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0029] In the industrial transportation sector, especially in special scenarios such as waste disposal and mining, vehicles are required to frequently perform precise unloading operations. Unloading operations typically require vehicles to park at specific locations next to the material pile, maintaining a precise distance and angle from the pile to ensure safety and efficiency during the unloading process.

[0030] In the exemplary technology, the driver determines the unloading location of the material pile, and then the driver drives the vehicle to the unloading location to unload the material.

[0031] However, drivers determine the unloading position of the material pile based on their own experience. Inexperienced drivers need to repeatedly drive the vehicle to adjust the position, resulting in a long unloading time and low vehicle efficiency.

[0032] In view of this, the embodiments of this application aim to provide a vehicle unloading method and related equipment. When a vehicle has an unloading requirement, point cloud data obtained by the vehicle's LiDAR scanning the environment around the vehicle and the vehicle's pose data are acquired. The priority of each unloading area in the stockpile where the vehicle is located is determined. Based on the priority, point cloud data, pose data, and prediction model, the vehicle's driving trajectory in the stockpile is predicted, and the vehicle is controlled to travel along the driving trajectory to the stockpile at the end of the driving trajectory for unloading. In this application, by using the point cloud data of the environment scanned by the vehicle's LiDAR, the vehicle's pose data, the priority of each stockpile area in the stockpile, and the prediction model, the unloading driving trajectory for the vehicle is automatically and accurately planned, enabling the vehicle to accurately and quickly reach the stockpile for unloading based on the driving trajectory, thereby improving the vehicle's unloading efficiency.

[0033] The vehicle unloading method and related equipment provided in this application embodiment can be applied to vehicle material stacking scenarios.

[0034] Exemplary methods Figure 1 A flowchart of a vehicle unloading method provided in this application embodiment Figure 1 .like Figure 1 As shown, the vehicle unloading method provided in this embodiment includes: Step S101: When the vehicle has a need to unload, acquire point cloud data and vehicle pose data obtained by the LiDAR in the vehicle scanning the environment where the vehicle is located.

[0035] In this embodiment, the vehicle is equipped with a positioning module. The vehicle obtains its current position in real time through the positioning module. If the current position is in the material storage yard and the weight sensor in the vehicle detects that its weight indicates it is under load, then it is determined that the vehicle has an unloading requirement. Alternatively, if a command to activate the unloading function is detected, it can also be considered that the vehicle has an unloading requirement.

[0036] The vehicle is equipped with a lidar system to detect the environment in which the vehicle is located and obtain point cloud data. The vehicle is also equipped with a pose sensor to detect the vehicle's pose data.

[0037] When a vehicle needs to unload, it acquires point cloud data collected by lidar and vehicle pose data detected by pose sensors.

[0038] Step S102: Determine the priority of each unloading area in the stockpile where the vehicle is located.

[0039] After obtaining the pose data, the vehicle determines the priority of each unloading area in the stockpile where it is located. Specifically, the stockpile includes multiple unloading areas, each of which can unload material; that is, each unloading area includes at least one stockpile. The vehicle can set priorities based on the parameters of the stockpiles in the unloading area. For example, the density, vehicle accessibility parameters, and stability parameters of the material pile in the unloading area are obtained. The density of the material pile can be determined by its material; for example, if the pile is sand or mud, the density is higher, and if it is waste kitchen waste, the density is lower. The vehicle accessibility parameter refers to whether a vehicle can reach the material pile. If it can, the accessibility parameter is 1; if it cannot, the accessibility parameter is 0. The stability parameter of the material pile is determined based on its weight, density, and height. The lower the weight, the lower the density, and the higher the height of the pile, the lower the stability parameter and the greater the safety hazard. Vehicles are assigned scores for density, accessibility, and stability parameters respectively. The sum of these scores determines the priority of the unloading area; the higher the sum of the scores, the higher the priority. Higher density, accessibility, and stability parameters result in higher scores for each parameter.

[0040] Step S103: Based on point cloud data, pose data, priority, and the prediction model in the vehicle, predict the vehicle's trajectory in the stockpile.

[0041] After obtaining the priorities, the vehicle's trajectory in the stockpile is predicted based on the point cloud data, pose data, priorities, and vehicle settings.

[0042] The prediction model learns the distribution of material piles in various unloading areas of the stockpile. By inputting point cloud data, pose data, and priorities into the prediction model, it can predict the optimal unloading position for the vehicle. Based on the environment represented by the point cloud data, the vehicle shape and driving status represented by the pose data, and combined with the optimal unloading point and the priorities of each unloading area, the model plans the vehicle's trajectory to the optimal unloading point. The prediction model can be a hybrid model of convolutional neural network and deep learning network.

[0043] The vehicle can train a predictive model. For example, the vehicle acquires various training samples, which include information such as stockpile maps of multiple stockpile areas and the priority of these areas. The stockpile maps are drawn based on point cloud data collected by LiDAR and can be grayscale images. Additionally, the training samples may include the vehicle's positioning data and sensor configuration files. The priority of the stockpile areas can be configured manually, similar to setting the stockpile density, vehicle accessibility parameters, and stockpile stability parameters for each area.

[0044] In addition, the training data also includes discretized trajectory points from point cloud data, that is, the trajectory points of the point cloud data are discretized into tokens, where token_nums=100 represents 100 location points. The prediction model structure includes an input layer, where the input dimension can be 256×256×2; the input layer is connected to a convolutional layer, where the convolutional kernel can be 7×7, and the dimension of the features input to the convolutional layer can be 128×128×64; the convolutional layer is connected to a batch normalization layer, where the dimension of the features input to the batch normalization layer can be 128×128×64; the batch normalization layer is connected to an activation function layer, where the dimension of the features input to the activation function layer can be 128×128×64; the activation function layer is connected to a max pooling layer, where the dimension of the features input to the max pooling layer can be 64×64×64; the activation function layer is connected to the residual stage 1 layer, where the dimension of the features input to the residual stage 1 layer can be 64×64×64. The dimension can be 64×64×64; Residual stage 1 connects to residual stage 2, and the dimension of the features input to residual stage 2 can be 32×32×128; Residual stage 2 connects to residual stage 3, and the dimension of the features input to residual stage 3 can be 16×16×256; Residual stage 3 connects to the data transformation layer, which converts multidimensional data into one-dimensional data; The data transformation layer connects to the deep learning encoder, the vehicle's current position is input to the masking module in the model, the output data of the masking module is input to the embedding layer, the output data of the embedding layer and the output data of the data transformation layer are input to the deep learning encoder, and the deep learning encoder outputs the driving trajectory.

[0045] After obtaining the training samples, the preset model is trained using these samples to obtain the prediction model. Data is manually labeled (priority of material pile areas) to ensure that the predicted vehicle trajectory logic aligns with the driver's vehicle operation logic, thus preventing the predicted trajectory from deviating from the driver's normal driving path.

[0046] It should be noted that various material stacking data can be collected in both simulated and real-world scenarios, such as metal stacking, construction material stacking, and kitchen waste stacking, to obtain training samples. These training samples cover a wide range of material stacking scenarios, thereby increasing the model's generalization ability. Furthermore, the prediction model built using a convolutional neural network model combined with a deep learning network model can adapt to unstructured input features, improving the accuracy of the prediction model in predicting vehicle trajectories.

[0047] Step S104: Control the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading.

[0048] After obtaining the driving trajectory, the end point of the driving trajectory is the unloading position. Control the vehicle to drive to the material pile at the end point of the driving trajectory for unloading.

[0049] In this embodiment, when a vehicle needs to unload, point cloud data obtained from the vehicle's LiDAR scanning of its environment and the vehicle's pose data are acquired. The priority of each unloading area in the stockpile where the vehicle is located is determined. Based on the priority, point cloud data, pose data, and a prediction model, the vehicle's trajectory in the stockpile is predicted, and the vehicle is controlled to travel along the trajectory to the stockpile at the end of the trajectory for unloading. In this application, by using point cloud data from the vehicle's LiDAR scanning environment, the vehicle's pose data, the priority of each stockpile area in the stockpile, and a prediction model, the unloading trajectory is automatically and accurately planned for the vehicle. This allows the vehicle to accurately and quickly reach the stockpile for unloading based on the trajectory, improving the vehicle's unloading efficiency.

[0050] Figure 2 A flowchart of a vehicle unloading method provided in this application embodiment Figure 2 ,based on Figure 1 In the embodiment shown, step S103 includes: Step S201: Correct the point cloud data based on the pose data to obtain intermediate point cloud data, and extract the height of each material pile from the intermediate point cloud data.

[0051] In this embodiment, after the vehicle obtains the point cloud data, since the coordinates of the point cloud data are the coordinates of the lidar, it is necessary to first convert the coordinates of the point cloud data into coordinates in the vehicle coordinate system.

[0052] Specifically, the point cloud data is first converted from the coordinates of the LiDAR to the coordinates of the vehicle's coordinate system. Then, the LiDAR extrinsic parameters are used to transform the point cloud data to the vehicle's coordinate system. Finally, the coordinate transformation from the LiDAR coordinates to the vehicle's coordinate system is performed. .

[0053] in, The coordinates of a point in the lidar coordinate system are typically a three-dimensional vector (x, y, z) that describes the point's position in the lidar coordinate system. : is a 3×3 rotation matrix that represents the rotation transformation from the LiDAR coordinate system to the vehicle coordinate system. It describes the attitude (i.e. orientation) of the LiDAR coordinate system relative to the vehicle coordinate system, and transforms the vectors in the LiDAR coordinate system to the vehicle coordinate system through rotation. : is a 3×1 translation vector representing the translation transformation from the lidar coordinate system to the vehicle coordinate system. It describes the position offset of the lidar coordinate system origin in the vehicle coordinate system. : is the coordinate of the transformed point in the vehicle coordinate system, which is also a three-dimensional vector (x′, y′, z′).

[0054] During LiDAR scanning, the vehicle may be in motion, such as accelerating or turning, causing discrepancies between the actual and theoretical positions of different points in the point cloud. To address this, after transforming the point cloud data into coordinates, pose data is used to correct the point cloud data; that is, pose data is used to compensate for motion distortion in the point cloud. Specifically, the vehicle position T(t) and pose R(t) at time t for each point in the transformed point cloud data can be interpolated using pose data, thereby correcting the distortion caused by motion.

[0055] In addition, before correcting the point cloud based on the pose, it is necessary to convert the coordinates of the point cloud in the vehicle coordinate system to the coordinates in the world coordinate system based on the pose data.

[0056] For example, vehicle pose The following conversion: .

[0057] : Represents the coordinates of a point in the point cloud in the vehicle coordinate system, usually a three-dimensional vector (x, y, z); : is a 3×3 rotation matrix that represents the rotation transformation from the vehicle coordinate system to the global coordinate system. It describes the vehicle's attitude (i.e., orientation) relative to the global coordinate system in the positive direction. It transforms the vectors in the vehicle coordinate system to the global coordinate system through rotation. : is a 3×1 translation vector, representing the translation transformation from the vehicle coordinate system to the global coordinate system. It describes the position offset of the origin of the vehicle coordinate system in the global coordinate system. : This refers to the coordinates of the transformed point in the global coordinate system, which is also a three-dimensional vector (x′, y′, z′). After completing the transformation of the point cloud into the world coordinate system, motion compensation is performed on the point cloud based on the pose data to obtain the intermediate point cloud data.

[0058] After obtaining the intermediate point cloud data, the stacking height is extracted from the intermediate point cloud data to obtain the stacking height of each item. The stacking height is as follows: in, The value of the Z-axis in the world coordinate system of the point cloud data. The value of the X-axis in the world coordinate system of the point cloud data; This represents the Y-axis value in the world coordinate system of the point cloud data. The material pile height is actually the grid height of the point cloud data.

[0059] Step S202: Construct a height matrix based on the height of each extracted material pile, and construct a priority matrix based on each priority level.

[0060] After obtaining the height of each material pile, a height matrix is ​​constructed based on the height of each material pile.

[0061] In one example, the height matrix can be obtained by constructing a matrix from the heights of each stack of materials. .

[0062] In another example, the matrix to be processed can be obtained by constructing matrices from the heights of each stack. By sequentially performing time-domain filtering and spatial-domain denoising on the matrix to be processed, the height matrix can be obtained. For example, time-domain filtering: for continuous... Frame height matrix Perform a moving average: Spatial domain noise reduction: Median filter kernel (3×3), i.e. .

[0063] After obtaining the height matrix, a priority matrix is ​​constructed based on the priority of each stockpile. For example, the priority weights for the unloading areas are... : Where x, y, z are predefined gradient data. .

[0064] Step S203: The height matrix and priority matrix are fused to obtain a fused matrix, and the fused matrix is ​​input into the prediction model to obtain the vehicle's driving trajectory in the stockpile.

[0065] After obtaining the height matrix and priority matrix, they are fused to obtain the fused matrix. The fused matrix is ​​as follows: .

[0066] The vehicle inputs the fusion matrix into the prediction model, and the prediction model outputs the predicted driving trajectory based on the fusion matrix.

[0067] For example, the input tensor X (fusion matrix) is processed by the residual network of the prediction model to obtain the final feature F (256×256): The encoder of the deep learning network in the prediction model is based on Prediction is performed: the feature interactions of the initially discretized feature points are... Through multiple iterations of prediction (current TokenNum=100), Traj is the final output trajectory, q0 is the initial token, currently set to

[100] , and F is the encoder operation for X.

[0068] In this embodiment, the shape recognition of the stockpile is replaced by the height of the point cloud (grid height or stockpile height). It does not rely on the geometry of the stockpile, but determines the drivable area and unloading area of ​​the vehicle in the stockpile based on the height change of the stockpile.

[0069] In this embodiment, point cloud data is corrected using pose data to obtain intermediate point cloud data, and the height of each material pile is extracted from the point cloud data to obtain the height of each material pile. A height matrix is ​​then constructed based on the extracted material pile heights, and a priority matrix is ​​constructed based on each material pile height. The priority matrix and the height matrix are then fused to obtain a fusion matrix. Finally, the fusion matrix is ​​input into the prediction model, enabling the prediction model to plan an accurate driving trajectory for vehicle unloading based on the fusion matrix.

[0070] Figure 3 A flowchart of a vehicle unloading method provided in this application embodiment Figure 3 .based on Figure 2 In the embodiment shown, step S201 includes: Step S301: Obtain obstacle data of the vehicle's environment collected by the millimeter-wave radar in the vehicle, and filter the point cloud data for dynamic obstacles based on the obstacle data to obtain processed point cloud data. The processed point cloud data includes point clouds corresponding to multiple material piles.

[0071] In this embodiment, a millimeter-wave radar is installed in the vehicle to collect obstacle data of the vehicle's environment. The millimeter-wave radar can penetrate dust to identify obstacles. The obstacle data collected by the millimeter-wave radar can be fused with the obstacle data represented in the point cloud data collected by the lidar to compensate for the blind spots of the lidar.

[0072] Specifically, the device acquires obstacle data of the vehicle's environment collected by millimeter-wave radar in the vehicle, and filters the point cloud data for dynamic obstacles based on the obstacle data to obtain a filtered point cloud. For example, the obstacle data contains both dynamic and static obstacles. Dynamic obstacles include moving vehicles, while static obstacles are like stockpiles of material in a storage yard. The point cloud data also contains multiple obstacles. Based on the obstacle data, dynamic obstacles in the point cloud data can be identified, and the point clouds representing dynamic obstacles are deleted. Furthermore, the static obstacles in the obstacle data are compared one-to-one with the static obstacles in the point cloud data. If the obstacle data contains a static obstacle not present in the point cloud data, a point cloud corresponding to the target obstacle is generated in the point cloud data. The target obstacle is the static obstacle not present in the point cloud data. The coordinates of the point cloud corresponding to the target obstacle are determined based on the coordinates of the target obstacle in the obstacle data. This process yields the processed point cloud data.

[0073] Furthermore, removing points representing dynamic obstacles from the point cloud data can also remove points representing dust, resulting in processed point cloud data. Points representing dust appear as isolated points in the point cloud data, meaning that the distance between these points and other points is greater than a distance threshold.

[0074] Step S302: Correct the point cloud of the processed point cloud data based on the pose data.

[0075] After obtaining the processed point cloud data, it can be corrected based on the pose data. For details on the process of correcting the processed point cloud data based on the pose data, please refer to [link / reference needed]. Figure 2 The embodiments shown will not be described in detail here.

[0076] In this embodiment, obstacle data collected by millimeter-wave radar is fused with point cloud data collected by lidar to remove dynamic obstacles from the point cloud data, thereby reducing the workload of the prediction model in predicting vehicle trajectories and improving the prediction efficiency of the prediction model.

[0077] Figure 4 A flowchart of a vehicle unloading method provided in this application embodiment Figure 4 .based on Figures 1 to 3 In any of the embodiments shown, step S104 includes: Step S401: Determine the driving direction of the vehicle at each path point on the driving trajectory and the target direction corresponding to each path point.

[0078] In this embodiment, after obtaining the driving trajectory, the driving trajectory can be decomposed into multiple sub-trajectories, so that the driver can drive based on each sub-trajectory segment, enabling the driver to drive the vehicle to the end of the driving trajectory more effectively.

[0079] For example, based on the driving trajectory, the driving direction of the vehicle at each path point on the trajectory and the target direction of each path point can be determined. The driving direction refers to the direction the vehicle is heading, and the target direction refers to the direction the path point is facing. The target direction can be the direction of the tangent line of the driving trajectory where the path point is located. The driving direction can be determined by simulating the direction the vehicle travels along the driving trajectory.

[0080] Step S402: Based on the driving direction and target direction corresponding to each path point, the driving trajectory is divided into multiple sub-trajectories. The directional angle between the target directions of adjacent path points on the sub-trajectories is less than a first directional angle threshold, and the directional angle between the driving directions of adjacent path points on the sub-trajectories is less than a second directional angle threshold.

[0081] After determining the driving direction and target direction, the driving trajectory can be divided into multiple sub-trajectories based on the driving direction and target direction corresponding to each waypoint. The directional angle between the target directions of adjacent waypoints on a sub-trajectory is less than a first angle threshold, and the directional angle between the driving directions of adjacent waypoints on a sub-trajectory is less than a second angle threshold. That is, the paths on the driving trajectory with roughly the same driving direction and roughly the same orientation of waypoints are considered as sub-trajectories.

[0082] Step S403: Control the vehicle to travel along each sub-track to the material pile at the end of the driving track for unloading. The vehicle speed gear remains unchanged while traveling on the sub-track.

[0083] After obtaining each sub-track, the vehicle can be controlled to travel along each sub-track to the material pile at the end of the driving track for unloading. When the vehicle is traveling on the sub-track, the vehicle speed gear remains unchanged. For example, the vehicle speed gear is 1st gear on sub-track 1 and 2nd gear on sub-track 2; or the vehicle speed gear is forward gear on sub-track 1 and reverse gear on sub-track 2.

[0084] In this embodiment, the vehicle divides its driving trajectory into multiple sub-trajectories, thereby enabling the driver to better navigate the vehicle to the end of the driving trajectory.

[0085] Figure 5 A flowchart of a vehicle unloading method provided in this application embodiment Figure 5 .based on Figures 1 to 4 In any of the embodiments shown, step S104 includes: Step S501: Obtain the kinematic constraints of the vehicle.

[0086] Step S502: Process the vehicle trajectory according to the kinematic constraints to obtain the target trajectory.

[0087] In this embodiment, the vehicle processes its driving trajectory so that the processed trajectory can adapt to the vehicle's actual driving needs.

[0088] In response, the vehicle acquires kinematic constraints, which are as follows: ; ; ; .

[0089] Vehicle trajectory processing using kinematic constraints can be achieved by minimizing an objective function, which is: ; in, The endpoint of the driving trajectory. These are the trajectory points on the driving path. For vehicle steering angle, To smooth the weights, the driving trajectory is smoothed using kinematic constraints to obtain the target trajectory. The target trajectory meets the vehicle's control requirements, such as comfort and drivability.

[0090] Step S503: Control the vehicle to travel along the target trajectory to the material pile at the end of the target trajectory for unloading.

[0091] After determining the target trajectory, control the vehicle to travel along the target trajectory to the material stacking point at the end of the target trajectory for unloading.

[0092] In this embodiment, the driving trajectory is processed by kinematic constraints so that the processed driving trajectory can meet the driver's needs for vehicle comfort and drivability.

[0093] The following is a general description of this application. The technical problem to be solved by this application is: 1. Parking positioning problem in unstructured environments: Current special operation scenarios lack fixed reference signs found in traditional parking environments; No fixed parking spaces: Traditional automatic parking systems rely on preset parking spaces or parking line markings, but industrial material storage scenarios completely lack such structured references; Dynamic reference frame: The shape, location, and volume of the waste pile change continuously over time, making it impossible to establish a static environment model. Lack of standard geometric features: Waste piles typically exhibit irregular and discontinuous geometric shapes, rendering traditional positioning methods based on geometric feature matching ineffective; Varied surface characteristics: The reflective properties of waste pile surfaces vary greatly (such as metal waste and construction waste), resulting in unstable sensor readings.

[0094] 2. Problem of determining the adaptive unloading position: The dynamic characteristics of waste disposal areas present unique challenges. Real-time stockpile status assessment: It is necessary to accurately identify the three-dimensional outline of the current waste pile and determine the optimal unloading position to avoid excessive local accumulation; Unloading point optimization: The working range of the unloading machinery, vehicle stability, and overall balance of the waste pile must be considered. Dynamic zone division: As the operation progresses, the effective unloading area changes continuously, and the system needs to re-evaluate it in real time; Multi-objective optimization: It is necessary to simultaneously satisfy mutually restrictive objectives such as maximizing unloading efficiency, vehicle safety, and uniform waste distribution.

[0095] 3. Mobility issues in confined spaces: The reduction in workspace due to waste accumulation has created special needs; Progressive space encroachment: As unloading operations proceed, the available operating space shrinks non-linearly; Real-time obstacle avoidance planning: The drivable area model must be dynamically updated to avoid collisions with newly added waste piles; Complex maneuver constraints: The turning radius and rear suspension sway of heavy vehicles in confined spaces must be taken into account in path planning.

[0096] 4. The problem of reliable identification of non-standard objects: The unique properties of waste piles render traditional identification methods ineffective. Morphological uncertainty: The waste pile lacks stable geometric features, making it impossible to apply template matching-based identification methods; Material diversity: The reflective properties of different waste materials (metal, plastic, construction waste, etc.) vary greatly, resulting in poor consistency of sensor data; Surface discontinuities: Waste piles typically contain numerous pores and abrupt edges, causing breaks in the 3D point cloud data; Dynamic deformation characteristics: During the unloading process, the waste pile may collapse or slide, forming a time-varying topology; Environmental interference: Dust, dirt and other interference factors in industrial settings further reduce the reliability of traditional vision systems.

[0097] To address the aforementioned issues, a vehicle unloading method is proposed. Its innovation lies in integrating multi-sensor data, constructing an adaptive unloading decision model, and combining deep learning and motion planning to solve the problem of automatic parking and unloading in industrial waste stacking scenarios.

[0098] Specific innovations include: Multi-frame point cloud fusion (e.g., for continuous) Frame height matrix Moving average and height feature extraction: By fusing multi-frame LiDAR point cloud data, stable height information is extracted to overcome the problem of inconsistent sensor data caused by differences in waste material.

[0099] Multidimensional fusion 2D map construction (fusion matrix): Combining lidar height information with unloading priority (such as stockpile density, drivable area, etc.), a fused multidimensional depth map is generated to provide structured input for decision-making.

[0100] Trajectory prediction based on neural networks: A convolutional neural network + deep learning network architecture is used to extract and decode features from the fused depth map to generate an anthropomorphic unloading trajectory.

[0101] Data-driven human-like training: Using manual loading and unloading data for model training makes autonomous driving strategies more in line with human operating habits and improves adaptability.

[0102] Kinematic optimization trajectory planning: Combining vehicle dynamics constraints, the trajectory output by the neural network is optimized to ensure that the vehicle can drive smoothly and avoid dynamic and static obstacles.

[0103] Furthermore, the solutions to the aforementioned technical problems are as follows: (1) Parking positioning problem in unstructured environments Technical problem: There are no fixed parking spaces in the waste storage area, and the waste is irregularly shaped, rendering traditional visual recognition methods ineffective.

[0104] Solution: Multi-frame point cloud fusion: By fusing point cloud data over time, noise (such as waste reflection and dust interference) is filtered out, stable height information is extracted, and a 3D environment model is constructed.

[0105] Height feature extraction: Calculate the height gradient of the waste pile based on point cloud data to identify the drivable area (low height zone) and the unloading area (high height zone).

[0106] Dynamic reference frame establishment: By utilizing synchronous positioning and mapping technology, combined with pose data and wheel velocity data, high-precision positioning can be achieved in environments without fixed landmarks.

[0107] Innovative Application: Multi-frame point cloud fusion → Solving the problem of unstable sensor data caused by waste materials. Multidimensional fusion 2D graph (fusion matrix) → Combines height information with priority information to form a structured input.

[0108] (2) Problem of determining the adaptive unloading position Technical issue: The shape of the waste pile changes dynamically, requiring real-time calculation of the optimal unloading point to avoid excessive local accumulation or vehicles getting stuck.

[0109] Solution: Multi-dimensional fusion 2D map: Combines the height information of LiDAR with unloading priorities (such as stockpile density, vehicle accessibility, and stability) to generate a height map and identify the optimal unloading area.

[0110] Convolutional Neural Network + Deep Learning Network Feature Extraction: The neural network learns the material distribution pattern from the fused map and predicts the optimal unloading point, avoiding the limitations of manual rule calculation. The convolutional neural network uses RestNet as the backbone network and adjusts the input layer parameters of the network.

[0111] Training on a dataset of manual tasks: The model is trained by collecting a dataset of manual tasks, so that the system can choose unloading strategies like a skilled driver (such as filling low-lying areas first and then expanding outwards).

[0112] Innovation Application: Multi-dimensional fusion of 2D graphs → provides structured decision input; Convolutional neural network + deep learning network → enables intelligent unloading point selection; Human-trained data training → Makes the model more in line with human operational logic.

[0113] (3) Mobility problem in confined space Technical problem: As waste accumulates, the space available for vehicles to travel is compressed, requiring multiple adjustments to their positions, which traditional path planning methods cannot adapt to.

[0114] Solution: Human-like trajectory planning: Through a large amount of training data from manual unloading, various material pile scenarios can be covered. The trajectory output by the neural network is decomposed into multiple sub-targets based on the vehicle's orientation and the orientation of the path points. Each sub-target contains information about the same gear (forward / reverse). The vehicle executes step by step according to the gear information.

[0115] Kinematic optimization (smoothing the predicted trajectory with kinematic constraints): Combining the vehicle's minimum turning radius and rear suspension sway constraints, the trajectory is smoothed to ensure feasibility.

[0116] Dynamic obstacle avoidance (removing dynamic obstacles from obstacle data): The environment model is updated in real time to avoid newly added waste piles during the trajectory optimization stage.

[0117] Innovation application: Neural network trajectory prediction → generating preliminary feasible paths; Kinematic optimization (→ Ensure the trajectory conforms to vehicle dynamics to avoid jamming).

[0118] (4) The problem of reliable identification of non-standard objects Technical problem: Waste materials have no fixed shape, making them difficult for traditional visual networks to identify.

[0119] Solution: Point cloud height feature replacement for shape recognition: Instead of relying on the geometry of the waste, it determines the driving area and unloading area based on the height change rate.

[0120] Data augmentation training: Collect various waste pile data (metal, construction waste, etc.) in simulated and real-world scenarios to enhance the model's generalization ability.

[0121] Multi-sensor redundancy: Combining millimeter-wave radar (which penetrates dust) and ultrasound (for close-range detection) to compensate for the blind spots of lidar.

[0122] Application of innovative points: Multi-frame point cloud fusion → Improves data stability; Convolutional Neural Networks + Deep Learning Networks → Adapting to Unstructured Features.

[0123] Compared to existing technologies, this application has the following significant advantages in material stockpile scenarios: 1. Greater environmental adaptability Traditional methods rely on fixed markers (such as parking lines) or preset rules, which cannot adapt to the dynamic changes in waste piles.

[0124] This application overcomes problems such as waste reflection and dust interference by extracting stable height information through multi-frame point cloud fusion. It also dynamically analyzes unloading priorities using a multi-dimensional fused 2D image, adapting to waste piles of different materials and shapes.

[0125] 2. Smarter unloading decisions Traditional methods, which rely on manual rules or simple thresholds to determine unloading points, can easily lead to localized accumulation or vehicles getting stuck.

[0126] This application employs a predictive model to learn the optimal unloading strategy from the fused graph, achieving human-like decision-making. Training with human-generated data (Innovation Point 4) makes the model more closely resemble the operating logic of a skilled driver, improving efficiency and safety.

[0127] 3. Smoother trajectory planning Traditional methods: Traditional path planning tends to generate infeasible paths in narrow spaces, requiring repeated adjustments. It also fails to consider vehicle kinematic constraints, leading to execution failures.

[0128] After generating an initial trajectory using a neural network, this application employs kinematic optimization to ensure compliance with vehicle turning radius, rear suspension sway, and other limitations. It supports progressive adjustments to adapt to dynamically compressed driving space.

[0129] 4. Higher robustness and generalization ability Traditional methods: visual solutions are greatly affected by dust and light; single sensors (such as pure LiDAR) are prone to failure in complex scenes.

[0130] This application: Multi-sensor fusion (LiDAR + millimeter wave + pose sensor) improves data reliability; data-driven training enables the model to adapt to various materials such as metal and construction waste without the need to redesign rules.

[0131] Exemplary device Corresponding to the above-described vehicle unloading method, this application also provides a vehicle. Figure 6 This is a schematic diagram of a vehicle module provided in an embodiment of this application. The vehicle provided in this embodiment includes: The acquisition module 610 is used to acquire point cloud data and vehicle pose data obtained by the LiDAR in the vehicle scanning the environment where the vehicle is located when the vehicle has unloading requirements. Determine module 620 to determine the priority of each unloading area in the stockpile where the vehicle is located; The prediction module 630 is used to predict the vehicle's trajectory in the stockpile based on point cloud data, pose data, priority, and the prediction model in the vehicle. The control module 640 is used to control the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading.

[0132] In some implementations, vehicle 600 is also used for: The point cloud data is corrected based on the pose data to obtain intermediate point cloud data, and the height of each material pile is extracted from the intermediate point cloud data. Construct a height matrix based on the height of each extracted material pile, and construct a priority matrix based on each priority level; The height matrix and priority matrix are fused to obtain a fused matrix, which is then input into the prediction model to obtain the vehicle's trajectory in the stockpile.

[0133] In some implementations, vehicle 600 is also used for: Based on the height of each extracted material pile, construct the matrix to be processed; The matrix to be processed is subjected to time-domain filtering and spatial-domain noise reduction in sequence to obtain the height matrix.

[0134] In some implementations, vehicle 600 is also used for: Obstacle data of the vehicle's environment collected by millimeter-wave radar in the vehicle are acquired, and the point cloud data is dynamically filtered for obstacles based on the obstacle data to obtain processed point cloud data, which includes point clouds corresponding to multiple material piles. The point cloud data of the processed point cloud data is corrected based on the pose data.

[0135] In some implementations, vehicle 600 is also used for: Determine the vehicle's direction of travel at each path point on the driving trajectory and the target direction corresponding to each path point; Based on the driving direction and target direction corresponding to each path point, the driving trajectory is divided into multiple sub-trajectories. The directional angle between the target directions of adjacent path points on the sub-trajectory is less than the first directional angle threshold, and the directional angle between the driving directions of adjacent path points on the sub-trajectory is less than the second directional angle threshold. The vehicle is controlled to travel along each sub-track to the material pile at the end of the track for unloading. The vehicle speed and gear remain unchanged while traveling on the sub-track.

[0136] In some implementations, vehicle 600 is also used for: Obtain the kinematic constraints of the vehicle; Based on kinematic constraints, the vehicle trajectory is processed to obtain the target trajectory; Control the vehicle and drive it along the target trajectory to the material pile at the end of the target trajectory to unload the material.

[0137] In some implementations, vehicle 600 is also used for: Acquire various training samples, including a material pile map containing multiple material pile regions and the priority of each material pile region in the material pile map. The material pile map is drawn based on point cloud data collected by LiDAR. The preset model is trained based on each training sample to obtain the prediction model.

[0138] The vehicle provided in this embodiment belongs to the same concept as the vehicle unloading method provided in the above embodiments of this application. It can execute the vehicle unloading method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the vehicle unloading method. Technical details not described in detail in this embodiment can be found in the specific processing content of the vehicle unloading method provided in the above embodiments of this application, and will not be repeated here.

[0139] The functions implemented by the various modules in the vehicle can be implemented by the same or different processors, and this application embodiment does not limit this.

[0140] It should be understood that the modules in the above-described vehicle can be implemented by a processor calling software. For example, the system includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of each module of the device. The processor can be a general-purpose processor, such as a CPU or microprocessor, and the memory can be internal to the device or external to the system. Alternatively, the modules in the system can be implemented as hardware circuits. By designing the hardware circuits, some or all of the module functions can be implemented. The hardware circuit can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and by designing the logical relationships between the components within the circuit, some or all of the above module functions are implemented. In another implementation, the hardware circuit can be implemented using a PLD, such as an FPGA, which can include a large number of logic gates. The connection relationships between the logic gates are configured through configuration files, thereby implementing some or all of the above module functions. All modules of the above-described vehicle can be implemented entirely by a processor calling software, entirely by hardware circuits, or partially by a processor calling software with the remaining parts implemented by hardware circuits.

[0141] In this application embodiment, a processor is a circuit with signal processing capabilities. In one implementation, the processor can be a circuit with instruction reading and execution capabilities, such as a CPU, microprocessor, GPU, or DSP. In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits. These logical relationships are fixed or reconfigurable. For example, the processor may be a hardware circuit implemented as an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the processor loading instructions to implement the functions of some or all of the above modules. Furthermore, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, or DPU.

[0142] As can be seen, each module in the above vehicle can be one or more processors (or processing circuits) configured to implement the above methods, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0143] Furthermore, the modules in the above-mentioned vehicle can be integrated in whole or in part, or they can be implemented independently. In one implementation, these modules are integrated together and implemented in the form of a System-on-Chip (SoC). The SoC may include at least one processor for implementing any of the above methods or implementing the functions of the modules of the device. The at least one processor may be of different types, such as CPU and FPGA, CPU and artificial intelligence processor, CPU and GPU, etc.

[0144] Exemplary electronic devices This application provides another structural schematic diagram of an electronic device, see [link to schematic diagram]. Figure 7 As shown, the electronic device includes a memory 700 and a processor 710; wherein the memory 700 is connected to the processor 710 and is used to store programs; the processor 710 is used to implement the vehicle unloading method disclosed in any of the above embodiments by running the programs stored in the memory 700.

[0145] Specifically, the aforementioned electronic device may further include: a bus, a communication interface 720, an input device 730, and an output device 740. The electronic device may also include a data transceiver module, an image monitoring module, and a signal monitoring module.

[0146] The processor 710, memory 700, communication interface 720, input device 730, and output device 740 are interconnected via a bus. Among them: A bus can include a pathway for transmitting information between various components in an electronic device.

[0147] The processor 710 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0148] The processor 710 may include a main processor, as well as a baseband chip, modem, etc.

[0149] The memory 700 stores a program that executes the technical solution of this invention, and may also store an operating system and other key business functions. Specifically, the program may include program code, which includes computer operation instructions. More specifically, the memory 700 may include read-only memory (ROM), other types of static storage devices capable of storing static information and instructions, random access memory (RAM), other types of dynamic storage devices capable of storing information and instructions, disk storage, flash memory, etc.

[0150] Input device 730 may include a device for receiving data and information input by a user, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.

[0151] Output device 740 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0152] The communication interface 720 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0153] The processor 710 executes the program stored in the memory 700 and calls other devices, and can be used to implement each step of any of the vehicle unloading methods provided in the above embodiments of this application.

[0154] This application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in the memory through the data interface to execute the vehicle unloading method described in any of the above embodiments. For the specific processing procedure and its beneficial effects, please refer to the above embodiments of the vehicle unloading method.

[0155] Exemplary computer program products and storage media In addition to the methods and apparatus described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the vehicle unloading methods according to various embodiments of this application as described in any of the foregoing embodiments of this specification.

[0156] Computer program products can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the power device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0157] Furthermore, embodiments of this application may also be storage media storing computer programs, which are executed by a processor to perform the steps of the vehicle unloading method according to various embodiments of this application described in any of the above embodiments of this specification, specifically implementing the steps of the above vehicle unloading method.

[0158] For the foregoing method embodiments, in order to simplify the description, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0159] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For apparatus embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0160] The steps in the methods of the various embodiments of this application can be adjusted, merged, or deleted in order according to actual needs, and the technical features described in each embodiment can be replaced or combined.

[0161] The units of the apparatus in the various embodiments of this application can be merged, divided, and deleted according to actual needs.

[0162] It should be understood that the disclosed terminals, devices, and methods can be implemented in other ways, given the several embodiments provided in this application. For example, the terminal embodiments described above are merely illustrative. For instance, the division of modules or sub-modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple sub-modules or modules may be combined or integrated into another module, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0163] The modules or submodules described as separate components may or may not be physically separate. The components that constitute a module or submodule may or may not be physical modules or submodules; that is, they may be located in one place or distributed across multiple network modules or submodules. Some or all of the modules or submodules can be selected to achieve the purpose of this embodiment's solution, depending on actual needs.

[0164] Furthermore, the functional modules or sub-modules in the various embodiments of this application can be integrated into one processing module, or each module or sub-module can exist physically separately, or two or more modules or sub-modules can be integrated into one module. The integrated modules or sub-modules described above can be implemented in hardware or in the form of software functional modules or sub-modules.

[0165] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0166] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software unit executed by a processor, or a combination of both. The software unit can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0167] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0168] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for unloading materials from a vehicle, characterized in that, include: When the vehicle has a need to unload, the point cloud data obtained by the lidar in the vehicle scanning the environment where the vehicle is located and the pose data of the vehicle are acquired. Determine the priority of each unloading area in the stockpile where the vehicle is located; Based on the point cloud data, the pose data, the priority, and the prediction model in the vehicle, the driving trajectory of the vehicle in the stockpile is predicted. Control the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading.

2. The vehicle unloading method according to claim 1, characterized in that, The step of predicting the vehicle's trajectory in the stockpile based on the point cloud data, the pose data, the priority, and the prediction model in the vehicle includes: The point cloud data is corrected based on the pose data to obtain intermediate point cloud data, and the height of each material pile is extracted from the intermediate point cloud data. A height matrix is ​​constructed based on the height of each extracted material pile, and a priority matrix is ​​constructed based on each of the aforementioned priorities; The height matrix and the priority matrix are fused to obtain a fusion matrix, and the fusion matrix is ​​input into the prediction model to obtain the vehicle's driving trajectory in the stockpile.

3. The vehicle unloading method according to claim 2, characterized in that, The construction of a height matrix based on the height of each extracted material pile includes: Based on the height of each extracted material pile, construct the matrix to be processed; The matrix to be processed is subjected to time-domain filtering and spatial-domain noise reduction in sequence to obtain the height matrix.

4. The vehicle unloading method according to claim 2, characterized in that, The step of correcting the point cloud data based on the pose data includes: Obstacle data of the vehicle's environment collected by the millimeter-wave radar in the vehicle are obtained, and the point cloud data is dynamically filtered for obstacles based on the obstacle data to obtain processed point cloud data, which includes point clouds corresponding to multiple material piles. The point cloud of the processed point cloud data is corrected based on the pose data.

5. The vehicle unloading method according to claim 1, characterized in that, Controlling the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading includes: Determine the driving direction of the vehicle at each path point on the driving trajectory and the target direction corresponding to each path point; Based on the driving direction and target direction corresponding to each path point, the driving trajectory is divided into multiple sub-trajectories. The directional angle between the target directions of adjacent path points on the sub-trajectories is less than a first directional angle threshold, and the directional angle between the driving directions of adjacent path points on the sub-trajectories is less than a second directional angle threshold. The vehicle is controlled to travel along each of the sub-trajectories to the material pile at the end of the travel trajectory for unloading, wherein the vehicle speed and gear remain unchanged while traveling on the sub-trajectories.

6. The vehicle unloading method according to claim 1, characterized in that, Controlling the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading includes: Obtain the kinematic constraints of the vehicle; Based on the kinematic constraints, the vehicle trajectory is processed to obtain the target trajectory; Control the vehicle to travel along the target trajectory to the material pile at the end of the target trajectory for unloading.

7. The vehicle unloading method according to claim 1, characterized in that, Before acquiring the point cloud data obtained by the lidar in the vehicle scanning the environment where the vehicle is located and the vehicle's pose data, the method further includes: Acquire various training samples, including a material pile map containing multiple material pile regions and the priority of each material pile region in the material pile map, which is drawn based on point cloud data collected by LiDAR; The preset model is trained based on each of the training samples to obtain the prediction model.

8. A vehicle, characterized in that, include: The acquisition module is used to acquire point cloud data obtained by the lidar in the vehicle scanning the environment where the vehicle is located, as well as the vehicle's pose data, when the vehicle has a need to unload. The module determines the priority of each unloading area in the stockpile where the vehicle is located; The prediction module is used to predict the driving trajectory of the vehicle in the stockpile based on the point cloud data, the pose data, the priority, and the prediction model in the vehicle. The control module is used to control the vehicle to travel along the driving trajectory to the material pile at the end of the driving trajectory for unloading.

9. An electronic device, characterized in that, Including memory and processor, among which, The memory is connected to the processor and is used to store programs; The processor is used to implement the vehicle unloading method as described in any one of claims 1-7 by running the program in the memory.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the vehicle unloading method as described in any one of claims 1-7.

Images