Control method and system for autonomous unloading of carry-scraper

Abstract

The invention provides an autonomous unloading control method and system for a carry-scraper, and relates to the technical field of underground carry-scrapers, and the method comprises the steps: collecting roadway environment point cloud data; according to the roadway environment point cloud data, the position and posture of the carry-scraper are calculated; performing path control on the carry-scraper according to the pose of the carry-scraper in combination with the kinematics model of the articulated vehicle and the MPPI algorithm model; according to the target position, environment data of the dump truck are collected, and characteristics of the dump truck are extracted; fusing the characteristics of the dump truck to obtain fused characteristics of the dump truck, performing pose detection on the dump truck, and outputting coordinates of key points; according to the coordinates of the key points, the 3D pose of the dumper container is calculated; according to the 3D poses, key parameters of containers of the carry-scraper and the dumper are calculated; dynamically adjusting the posture of the bucket according to the key parameters; according to the adjusted posture of the bucket, global coordinates of bucket teeth of the bucket are calculated; and according to the global coordinates, the bucket teeth are adjusted so as to control the carry-scraper to carry out autonomous unloading.

Description

Technical Field

[0001] The present invention relates to the technical field of underground scrapers, and in particular to a control method and system for autonomous unloading of scrapers. Background Art

[0002] A scraper is a heavy machine used in mines, construction sites, and other locations, primarily for loading, transporting, and unloading materials such as ore and earth. Autonomous unloading control involves using computer control systems, sensors, and algorithms in unmanned or automated scrapers to enable them to complete unloading tasks autonomously.

[0003] With the development trend of electric and intelligent scrapers, people are committed to the fully automated operation of scrapers in the loading, transportation and unloading processes of ore. Therefore, controlling the scraper to achieve autonomous unloading control of materials is of great significance for reducing driving distances, reducing energy consumption, improving work efficiency, reducing human errors and reducing operating costs.

[0004] However, while the transport portion of the current loading-haul-unloading cycle can be executed autonomously, the unloading of the ore has not yet been fully autonomous. This involves the coordination between the loader and the dump truck, leading to unnecessary errors and inefficiencies, which increases labor costs and limits the efficiency and automation level of the loader in the ore unloading process. Summary of the Invention

[0005] In order to solve the technical problem that the transportation part of the current loading-transport-unloading cycle can realize the autonomous execution mode, but the unloading part of the ore cannot fully realize autonomous operation, which involves the coordination between the shovel loader and the dump truck, resulting in unnecessary errors and inefficiency, thereby increasing labor costs and limiting the efficiency and automation level of the shovel loader in the ore unloading process, the present invention provides a control method and system for autonomous unloading of the shovel loader.

[0006] The technical solutions provided by the embodiments of the present invention are as follows:

[0007] First aspect:

[0008] An embodiment of the present invention provides a method for controlling autonomous unloading of a scraper, comprising:

[0009] S1: Collecting point cloud data of roadway environment;

[0010] S2: Based on the roadway environment point cloud data, the position of the scraper is calculated using the nine-axis inertial measurement unit;

[0011] S3: Based on the position of the scraper, the scraper is controlled by combining the kinematic model of the articulated truck and the MPPI algorithm model to reach the target position.

[0012] S4: Based on the target location, environmental information data of the dump truck is collected, and features of the environmental information data are extracted through a neural network to determine features of the dump truck, wherein the environmental information data includes lidar point cloud data and RGB image data;

[0013] S5: The extracted dump truck features are fused through the cross-feature fusion block to obtain the dump truck fusion features. Based on the dump truck fusion features, YOLOv7-tiny is used as the architecture to detect the dump truck pose and output the key point coordinates of the dump truck cargo area.

[0014] S6: Calculate the 3D pose of the dump truck container using the EPnP algorithm based on the key point coordinates and the LiDAR point cloud data;

[0015] S7: Based on the 3D pose, the key parameters describing the relative position of the scraper and the dump truck cargo box are calculated through a dynamic unloading strategy;

[0016] S8: Combining key parameters and real-time environmental perception technology, it dynamically adjusts the scraper bucket's posture to prevent the bucket from rubbing against the dump truck's cargo box.

[0017] S9: Calculate the global coordinates of the bucket teeth using the vehicle model according to the adjusted bucket posture;

[0018] S10: According to the global coordinates, the bucket teeth of the scraper are adjusted to control the scraper to autonomously unload onto the dump truck.

[0019] Second aspect:

[0020] An embodiment of the present invention provides a control system for autonomous unloading of a scraper, comprising:

[0021] processor;

[0022] The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the control method for autonomous unloading of the scraper loader according to the first aspect is implemented.

[0023] The third aspect:

[0024] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the control method for autonomous unloading of a scraper loader according to the first aspect is implemented.

[0025] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0026] In an embodiment of the present invention, the pose of a scraper is generated by calculating the collected point cloud data of the roadway environment. The scraper is then path-controlled to reach the target location by combining the kinematic model of the articulated truck and the MPPI algorithm model. Based on the target location, environmental information data of the dump truck is collected, and features of the dump truck are extracted using a neural network. The extracted features are fused together. Based on the fused features of the dump truck, the pose of the dump truck is detected using the YOLOv7-tiny architecture, and the coordinates of the key points of the dump truck's cargo area are output. Based on the key point coordinates and the lidar point cloud data, the EPnP algorithm is used to calculate the 3D pose of the dump truck's cargo area. Based on the 3D pose, a dynamic unloading strategy is used to calculate key parameters describing the relative position of the scraper and the dump truck's cargo area. By combining key parameters and real-time environmental perception technology, the bucket posture of the scraper is dynamically adjusted to prevent the bucket from scratching the dump truck cargo box. According to the adjusted bucket posture, the global coordinates of the bucket teeth are calculated through the vehicle model. Based on the global coordinates, the bucket teeth of the scraper are adjusted to control the scraper to autonomously unload onto the dump truck. This achieves fully autonomous control, efficient path planning, accurate posture estimation and dynamic adjustment, reducing human intervention and improving work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 A schematic flow chart of a method for controlling autonomous unloading of a scraper provided by an embodiment of the present invention;

[0029] Figure 2 A schematic diagram of a process for a scraper to retreat to the main tunnel area after excavation is completed, provided by an embodiment of the present invention;

[0030] Figure 3 A schematic diagram of a process flow of a scraper moving from a main lane to a dump truck and completing unloading preparations provided by an embodiment of the present invention;

[0031] Figure 4 A schematic diagram of a structure for partitioning a cargo box based on the capacity of a scraper bucket provided by an embodiment of the present invention;

[0032] Figure 5 A schematic diagram of the structure of an autonomous unloading scraper provided by an embodiment of the present invention;

[0033] Figure 6A schematic structural diagram of a control system for autonomous unloading of a scraper provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0035] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0036] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0037] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0038] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0039] Reference Manual Figure 1 , which shows a flow chart of a control method for autonomous unloading of a scraper provided by an embodiment of the present invention.

[0040] An embodiment of the present invention provides a method for controlling the autonomous unloading of a scraper loader. The method can be implemented by a control device for the autonomous unloading of the scraper loader, which can be a terminal or a server. The process flow of the control method for the autonomous unloading of the scraper loader can include the following steps:

[0041] S1: Collect point cloud data of the roadway environment.

[0042] The tunnel environment refers to underground mines or other confined spaces that may contain complex obstacles and changing terrain, requiring precise perception and modeling. Point cloud data is three-dimensional spatial data generated by LiDAR or other sensors, representing the surfaces of objects in the scene.

[0043] Specifically, the tunnel environment point cloud data was collected by a 16-line lidar (Velodyne VLP-16).

[0044] It should be noted that by collecting point cloud data in the tunnels, the system can fully understand the environmental conditions of the underground mining area, including the location of obstacles, the shape of the tunnels, etc., providing a basis for subsequent path planning and position correction.

[0045] Reference Manual Figure 2 , shows a schematic diagram of a process in which a scraper loader retreats to the main tunnel area after excavation is completed, provided by an embodiment of the present invention.

[0046] Figure 2 In an underground loader, after completing its loading task, it reverses at low speed out of the mining area's tunnel and stops at a fixed area near the main underground tunnel, pointing the bucket toward the dump truck. During this process, the dynamic movement of the ore pile and the loader causes variations in the withdrawal distance after each excavation, making it difficult to establish an accurate loader dynamics model. Unmanned loader systems rely on LiDAR to acquire information about their surroundings, sensing obstacles and boundary conditions in the tunnel in real time. Using the MPPI algorithm, they sample paths and select the optimal path planning solution by evaluating the costs of different paths, achieving simultaneous planning and control without a priori maps.

[0047] S2: Based on the point cloud data of the roadway environment, the position of the scraper is calculated using the nine-axis inertial measurement unit.

[0048] A nine-axis inertial measurement unit (IMU) is a sensor system used to measure an object's acceleration, angular velocity, and orientation. For a scraper, this position includes its coordinates in space (e.g., x, y, z) and orientation angles (e.g., heading, pitch, and roll).

[0049] It's important to note that the nine-axis IMU provides high-precision positioning information by inferring the LHD's position and posture in real time. The IMU can continuously track the LHD's position and posture changes in underground environments without external positioning systems (such as GPS). This process, independent of external environmental perception, offers strong real-time and stability, effectively addressing the complex and dynamic working environment of mining areas.

[0050] In one possible implementation, the calculation formula for calculating the position and posture of the scraper using the nine-axis inertial measurement unit in S2 is specifically:

[0051] v k+1 =v k +a k ·Δt

[0052] θ k+1 =θ k +ω z ·Δt

[0053] x k+1 =x k +v k cosθ k ·Δt

[0054] y k+1 =y k +v k sinθ k ·Δt

[0055] Among them, v k+1 represents the speed of the scraper at time k+1, v k represents the speed of the scraper at time k, a k represents the acceleration of the scraper at time k, Δt represents the control period, and θ k+1 represents the heading angle of the scraper at time k+1, θ k represents the heading angle of the scraper at time k, ω z represents the angular velocity of the scraper in the Z-axis (perpendicular to the ground), x k+1 represents the position of the scraper on the x-axis at time k+1, x k represents the position of the scraper on the x-axis at time k, and y k+1 represents the position of the scraper on the y-axis at time k+1, k represents the position of the scraper on the y-axis at time k.

[0056] It should be noted that based on ICP technology, the IMU and lidar data are fused through the extended Kalman filter (EKF) to correct the position of the scraper.

[0057] S3: Based on the position of the scraper, the scraper is controlled by combining the kinematic model of the articulated truck and the MPPI algorithm model to reach the target position.

[0058] The kinematic model of an articulated truck is a mathematical model that describes the vehicle's motion, including the relationships between physical quantities such as the vehicle's position, velocity, and acceleration. MPPI (Model Predictive Path Integral) is a model-based path planning algorithm that optimizes the vehicle's path by predicting multiple future time steps and selecting the optimal control sequence. The target location refers to the destination or designated area that the vehicle or scraper needs to reach.

[0059] It's important to note that efficient path control for the scraper is achieved by combining an articulated truck kinematic model with the MPPI algorithm. The articulated truck kinematic model accurately describes the scraper's motion in complex environments, especially when there's flexible steering between the front and rear vehicles. The MPPI algorithm, through real-time path optimization, ensures the scraper selects the optimal path in dynamic environments, avoiding obstacles and ensuring energy efficiency.

[0060] In a possible implementation, S3 specifically includes:

[0061] S301: Establishing the kinematic model of the articulated vehicle:

[0062]

[0063] X=[x f ,y f ,Φ f ,v f ,θ f ] T

[0064] u=[a,ω] T

[0065]

[0066] in, represents the kinematic model of the articulated truck, represents the state matrix, represents the input matrix, u represents the input vector, v f represents the speed of the front vehicle, Φ f Indicates the heading angle of the front vehicle, v r represents the speed of the rear vehicle, θ f Indicates the heading angle difference between the front and rear vehicles, l f Indicates the axle length of the front vehicle body, l r represents the axis length of the rear body, X represents the state vector, x f Indicates the horizontal coordinate of the front vehicle body, y f represents the ordinate of the front vehicle body, T represents the transposition, a represents the acceleration of the front vehicle body, ω represents the angular velocity of the articulation angle, Φ r Indicates the heading angle of the rear vehicle body.

[0067] S302: Combine the articulated vehicle kinematic model and the MPPI algorithm model to set the cost function:

[0068] c(x)=obstacle(x)+Speed(x)+Forward(x)

[0069]

[0070] r(t)=(x(t),y(t),z(t))

[0071] r 2D (t)=(x(t),y(t))=Proj xy (r(t))

[0072]

[0073] Forward(x)=-w*v x

[0074] Where c(x) represents the cost function, obstacle(x) represents the obstacle cost function, Speed(x) represents the speed cost function, Forward(x) represents the forward direction cost function, ξ1 and ξ2 represent weighting factors, exp represents the exponential function, mindist represents the minimum distance between the vehicle and the obstacle, hit represents a binary variable, and distance represents the difference between the vehicle coordinates (x, y) and the obstacle coordinates (obs x ,obs y ), r(t) represents the position vector of the vehicle in three-dimensional space at time t, x(t), y(t) and z(t) represent the position coordinates of the vehicle on the x, y and z axes at time t respectively, r 2D (t) represents the position vector of the vehicle on the two-dimensional plane at time t, Proj xy (r(t)) represents the projection of the three-dimensional position r(t) onto the xy plane, v 2D (t) represents the speed of the vehicle on the two-dimensional plane at time t, and Represent the speed components of the vehicle in the x and y directions respectively, Speed(t) represents the total speed of the vehicle at time t, λ represents the weighting factor, It represents the rate of change of the vehicle's speed on the two-dimensional plane, that is, acceleration, and ω represents the weighting factor.

[0075] S303: With the goal of minimizing the cost function value, the scraper is controlled to reach the target position.

[0076] The cost function is a mathematical function used to measure the quality of path selection. It quantifies the quality of each possible path during the loader's operation, helping the control system select the optimal path.

[0077] It should be noted that by establishing a cost function, the optimal control input sequence is determined to guide the vehicle from the initial state to the target position.

[0078] S4: According to the target position, environmental information data of the dump truck is collected, and features of the environmental information data are extracted through a neural network to determine features of the dump truck, wherein the environmental information data includes lidar point cloud data and RGB image data.

[0079] A dump truck is a vehicle commonly used to transport materials and has a cargo box that can automatically tilt. Environmental information data refers to data about the surrounding environment acquired through sensors (such as lidar and cameras). Neural networks are computational models that mimic the learning and reasoning of biological neural systems and are commonly used for tasks such as image recognition, classification, and regression.

[0080] LiDAR point cloud data is three-dimensional spatial data obtained through laser scanning, representing the spatial position of each point in the scene. RGB image data is image data obtained by a camera, which contains the color and texture information of the scene.

[0081] It should be noted that by combining LiDAR point cloud data and RGB image data, comprehensive environmental information about the dump truck is collected. This process can obtain multi-dimensional information such as the dump truck's three-dimensional spatial position, shape, and color, providing accurate data support for subsequent posture detection.

[0082] Specifically, the scraper uses a laser radar (LiDAR) and an RGB camera to simultaneously collect environmental information about the dump truck, obtaining the target's appearance and three-dimensional position. The RGB camera can provide texture information about the dump truck, such as the truck's edges and body color, but is significantly affected by lighting and shadows. LiDAR, on the other hand, generates sparse point cloud data, enabling precise measurement of the dump truck's spatial position and contours, even in low-light or bright-light environments.

[0083] In a possible implementation manner, after S4, the method further includes:

[0084] Preprocess the environmental information data. The preprocessing specifically includes:

[0085] Combine the rotation matrix and translation matrix to transform the lidar coordinates to the camera coordinates:

[0086]

[0087] Among them, [X C ,Y C ,Z C ] represents the coordinates in the camera coordinate system, R represents the rotation matrix, T represents the translation matrix, [X L ,Y L ,Z L ] represents the coordinates in the lidar coordinate system.

[0088] Project the transformed lidar coordinates into the camera's 2D image coordinate system:

[0089]

[0090] Among them, (x, y) represents the two-dimensional coordinates in the two-dimensional image coordinate system, f x Indicates the focal length of the camera on the x-axis, f y Indicates the focal length of the camera on the y-axis, c x Indicates the coordinate of the camera's optical center on the x-axis, c y Indicates the coordinate of the camera's optical center on the y-axis.

[0091] According to the two-dimensional image coordinate system, the lidar point cloud data is converted into a depth image with the same size as the RGB image data.

[0092] In the present invention, in order to improve the calculation efficiency of data, a method of converting point cloud data into depth image is adopted to project the three-dimensional point cloud data collected by LiDAR into the camera coordinate system to generate a two-dimensional depth image with the same size as the RGB image.

[0093] In a possible implementation, S4 is specifically:

[0094] The features of the dump truck are extracted from RGB image data and depth images through a neural network. The neural network contains two independent convolutional neural network branches with shared parameters. The convolutional neural network branches include: an RGB branch and a depth image branch. The RGB branch is responsible for extracting the texture features of the dump truck, and the depth image branch is responsible for extracting the geometric features of the dump truck.

[0095] Reference Manual Figure 3 , which shows a schematic diagram of a process of a scraper provided by an embodiment of the present invention moving from a main tunnel to a dump truck and completing unloading preparations.

[0096] Figure 3 In this paper, a fusion perception technology of LiDAR and RGB cameras is used, combined with a neural network and the Transformer attention mechanism for feature extraction and fusion. A dataset of cargo containers with different lighting, placement angles, shapes, and colors is constructed for network training. In addition, a method is proposed to dynamically partition the dump truck cargo container according to bucket volume, and adjust the bucket's unloading position based on the real-time load information of each partition.

[0097] S5: The extracted dump truck features are fused through the cross-feature fusion block to obtain the dump truck fusion features. Based on the dump truck fusion features, YOLOv7-tiny is used as the architecture to detect the dump truck pose and output the key point coordinates of the dump truck cargo area.

[0098] Among them, the cross feature fusion block is used to fuse features from different data sources (such as RGB images and depth images) to obtain a richer information representation. The dump truck feature refers to the visual and spatial information about the dump truck extracted from the RGB image and the depth image. The dump truck fusion feature is the result of fusing the feature information from the RGB image and the depth image through the cross feature fusion block. YOLOv7-tiny is a lightweight deep learning target detection network specially designed for fast and accurate object detection on resource-limited devices. Key point coordinates refer to the two-dimensional or three-dimensional coordinates output by the model during the dump truck detection process that represent the position of the dump truck cargo box or other important parts.

[0099] It's important to note that the cross-feature fusion block fuses the dump truck features extracted from the RGB and depth images, enabling the system to better capture the truck's color, shape, and spatial structure. After performing pose detection using YOLOv7-tiny, the system outputs the coordinates of key points in the dump truck's cargo area, providing accurate data for subsequent path planning and unloading control.

[0100] Specifically, during the object detection phase, we used YOLOv7-tiny as the base network. We optimized its last layer and replaced it with a pose regression head. This allows us to detect the dump truck while also outputting keypoint information about the cargo box. The YOLO network detects the corners of the cargo box and outputs their 2D coordinates. Low-confidence keypoints are filtered out by setting a confidence threshold.

[0101] In a possible implementation, the calculation formula of the dump truck fusion feature in S5 is specifically:

[0102] F fusion =W1·F RGB +W2·F Depth

[0103] Among them, F fusion represents the dump truck fusion feature, W1 and W2 both represent adaptive weight factors, F RGB represents the features extracted from RGB images, F Depth Represents features extracted from depth images.

[0104] It should be noted that the cross-feature fusion block is used to perform information interaction at the multi-scale convolutional level. The fusion method significantly improves the accuracy of dump truck detection and positioning, especially in complex environments.

[0105] In a possible implementation manner, after S5, the method further includes:

[0106] By introducing the Transformer-based self-attention mechanism, the attention to the dump truck cargo area is enhanced.

[0107] The calculation formula of Transformer's self-attention mechanism is as follows:

[0108]

[0109] Among them, A represents the attention weight matrix, softmax represents the normalization function, Q represents the query matrix, K represents the key matrix, V represents the value matrix, and d k represents the dimensions of the query and key vectors, and T represents the transpose.

[0110] It should be noted that since dump trucks may be affected by other equipment, walls, or lighting changes in the mining area, some target areas may be blocked or difficult to identify. Therefore, the Transformer-based attention mechanism is introduced to optimize the dump truck's pose estimation accuracy.

[0111] S6: Based on the key point coordinates and lidar point cloud data, the EPnP algorithm is used to calculate the 3D pose of the dump truck cargo box.

[0112] The EPnP algorithm is a computer vision algorithm used to recover the 3D pose (i.e., position and orientation) of an object from multiple 2D image points and their corresponding 3D world coordinates. 3D pose refers to the position and orientation of an object in 3D space.

[0113] It should be noted that the EPnP (Efficient Perspective-n-Point) algorithm is used to solve the 3D pose of the cargo box. This algorithm calculates the posture of the dump truck cargo box in space by comparing 2D key points with the corresponding 3D point cloud information. During the iterative optimization process, the maximum number of iterations is limited, while the reprojection error is controlled within 2 pixels to ensure real-time calculation and accuracy.

[0114] In a possible implementation manner, after S6, the method further includes:

[0115] The 3D pose is smoothed by Kalman filtering to reduce the impact of noise on the accuracy of pose detection.

[0116] Among them, Kalman filtering is a recursive filtering algorithm that is widely used in data estimation, state prediction, noise suppression and other problems in dynamic systems.

[0117] In the present invention, in order to improve the stability of pose estimation, a Kalman filter is introduced in the post-processing to smooth the pose data to reduce the jitter caused by sensor noise or environmental interference during the detection process.

[0118] S7: Based on the 3D pose, the key parameters describing the relative position of the scraper and the dump truck cargo box are calculated through a dynamic unloading strategy.

[0119] Dynamic unloading strategies adjust the movements and positions of the scraper and dump truck in real time based on the relative positions of the scraper and dump truck's cargo box and actual operational requirements. Key parameters are the key variables that describe the relative positions of the scraper and dump truck's cargo box during unloading. For example, the contact angle and distance between the scraper bucket and the cargo box directly impact the accuracy and efficiency of the unloading process.

[0120] It's important to note that the dynamic unloading strategy automatically adjusts the scraper's unloading process based on the real-time working environment and the relative position between the scraper and the dump truck. This strategy dynamically optimizes the bucket's posture, unloading path, and docking position with the dump truck's cargo box, avoiding spillage, collisions, and inefficient unloading caused by improper positioning.

[0121] S8: Combining key parameters and real-time environmental perception technology, the bucket posture of the scraper is dynamically adjusted to prevent the bucket from scratching the dump truck cargo box.

[0122] Real-time environmental perception technology refers to the use of sensors (such as lidar and cameras) to monitor the surrounding environment in real time. Bucket posture refers to the bucket's fixed position relative to the scraper or its angle to the ground, including its tilt, lift, and rotation.

[0123] It's important to note that real-time environmental awareness technology dynamically adjusts the scraper's bucket posture to ensure precise alignment with the dump truck's cargo box. The real-time feedback system allows for flexible adjustment of the bucket's posture based on varying unloading requirements or obstacles, preventing collisions with the cargo box or uneven unloading.

[0124] Reference Manual Figure 4 , showing a structural schematic diagram of a cargo box partition based on the scraper bucket volume provided by an embodiment of the present invention.

[0125] Figure 4 In the figure, according to the geometric shape and volume characteristics of the cargo box, the cargo box is divided into the front area (①), the middle front area (②), the middle rear area (③) and the top covering area (④).

[0126] This invention designs a dynamic unloading strategy based on the capacity of the scraper bucket and the loading characteristics of the dump truck's cargo box. Because the size and volume of dump truck cargo boxes vary depending on the vehicle model, this invention adopts a zoned unloading method. Based on the cargo box's geometry and volumetric characteristics, the cargo box is divided into a front area, a mid-front area, a mid-rear area, and a top cover area. The scraper first dumps the mineral material in the front, mid-front, and mid-rear areas of the cargo box to ensure that the cargo box is fully filled and the overall load distribution is more uniform. Finally, the top cover area is unloaded. After the bottom of the cargo box is filled, the scraper performs additional unloading in the top area of the cargo box, forming a uniform cover layer and optimizing the loading density.

[0127] S9: Calculate the global coordinates of the bucket teeth using the vehicle model according to the adjusted bucket posture.

[0128] The vehicle model is a mathematical model that describes the motion and posture of the scraper. The bucket teeth are the part of the scraper bucket used for digging and shoveling materials. Global coordinates are coordinates relative to the entire working environment or reference system.

[0129] It's important to note that by combining the vehicle model with real-time sensor data, the global coordinates of the bucket teeth are precisely calculated, ensuring that the scraper can precisely control their movement during unloading operations. This calculation of the global coordinates of the bucket teeth prevents collisions between the bucket and the dump truck's cargo box, ensuring accurate and efficient unloading.

[0130] The present invention calculates the position of the bucket teeth based on the vehicle model. The movement of the shovel arm directly affects the position of the bucket teeth, so it is necessary to obtain the lifting angle of the shovel arm. The present invention installs an angle sensor at the hinge point of the shovel arm to measure the lifting angle of the shovel arm in real time. In addition, the system arranges a roll angle sensor at the connection between the bucket and the shovel arm to measure the roll angle of the bucket. Finally, in order to establish the position model of the bucket teeth, the system needs to call the vehicle structural parameters of the scraper, including the position of the shovel arm rotation point and the fixed position of the bucket teeth in the bucket coordinate system. These parameters are determined by the design structure of the vehicle, calibrated during system initialization, and dynamically updated in combination with real-time measurement data during the unloading process.

[0131] In a possible implementation, the calculation formula of the global coordinates of the bucket teeth in S9 is specifically:

[0132]

[0133] in, represents the global coordinates of the bucket teeth, Indicates the center position of the scraper, R veh represents the vehicle posture rotation matrix, Indicates the position of the scraper arm rotation point, Rarm represents the rotation matrix of the lifting angle of the scraper arm, R bucket represents the scraper bucket flip angle matrix, Indicates the fixed position of the scraper bucket teeth in the bucket coordinate system.

[0134] Reference Manual Figure 5 , showing a structural schematic diagram of autonomous unloading of a scraper provided by an embodiment of the present invention.

[0135] Figure 5 In the system, the position and posture of the dump truck's cargo box are detected in real time through the fusion of LiDAR and an RGB camera. The global coordinates of the bucket teeth are then calculated based on the vehicle's kinematic model. The bucket is then controlled to reach the target height before the vehicle reaches the unloading point, based on its relative position to the cargo box. To ensure thorough unloading, the system further incorporates a bucket vibration control mechanism that generates small vibrations at a specific frequency, effectively dislodging any ore adhering to the bucket surface.

[0136] S10: According to the global coordinates, the bucket teeth of the scraper are adjusted to control the scraper to autonomously unload onto the dump truck.

[0137] Autonomous unloading refers to the automatic unloading process of the loader without human intervention. The implementation of autonomous unloading technology reduces human intervention and enables the loader to operate efficiently and stably in complex and dynamic mining environments.

[0138] In a possible implementation manner, after S10, the method further includes:

[0139] The MPPI algorithm model is used to control the path of the scraper until it is located in a fixed area of the main tunnel and the bucket is reset.

[0140] It should be noted that after the unloading task is completed, the scraper needs to reverse and return to the fixed area of the main tunnel. This system adopts real-time synchronous path planning and trajectory tracking control based on MPPI. The surrounding environment is reconstructed through laser radar (LiDAR) point cloud data to establish a real-time environment map. The MPPI path optimization module generates multiple candidate trajectories based on the current posture of the scraper and the target posture in the fixed area of the main tunnel, calculates the cost function, and selects the optimal reversing path. Path planning and tracking are performed simultaneously, that is, MPPI continuously updates the optimal reversing trajectory in each control cycle and adjusts the motion state of the scraper in real time. If a dynamic obstacle (such as personnel or other equipment) is detected, the system will immediately trigger local replanning and generate a new safe reversing path internally. When the scraper reverses and returns to the fixed area of the main tunnel, the system controls the bucket to reset synchronously to avoid additional waiting time and prevent the bucket from scratching the edge of the dump truck cargo box or other obstacles. First, during reverse, the system controls the gradual lowering of the bucket lift cylinder and simultaneously adjusts the retraction angle of the bucket tilt cylinder to ensure the bucket returns to its initial horizontal position before the LHD reaches a fixed area in the main roadway. Bucket lift control utilizes a linear height control model, dynamically adjusting the bucket's descent rate based on the LHD's reverse distance. Simultaneously, the tilt cylinder retracts along a preset trajectory, gradually restoring the bucket's tilt angle to a horizontal position. The system utilizes closed-loop PID control to ensure a smooth and impact-free bucket repositioning process, preventing vehicle instability caused by posture adjustments.

[0141] Secondly, the system uses LiDAR and IMU sensor fusion to calculate the minimum distance between the bucket and its surroundings. If the distance between the bucket and the edge of the cargo box is less than a set safety threshold, the system adjusts the bucket's lift height. If the vehicle's posture deviation exceeds a set threshold, the system pauses reverse and adjusts the bucket's posture to maintain a safe distance throughout the repositioning process.

[0142] Furthermore, because MPPI path planning and trajectory tracking are executed simultaneously, the system optimizes the loader's trajectory in real time during reverse and simultaneously adjusts the bucket's return trajectory, ensuring the bucket's posture remains consistent with the vehicle's reverse path. If the system detects path deviation or a risk of scratching, it dynamically adjusts the bucket's posture using an incremental position compensation algorithm. Ultimately, by the time the loader reaches a fixed area in the main roadway, the bucket has already been returned to its original position, eliminating the need for further adjustments and ensuring readiness for the next round of operations.

[0143] After completing the reverse reset, the scraper enters standby mode and prepares for the next cycle operation.

[0144] In the present invention, the cargo box point cloud data is acquired through LiDAR scanning, and the key points of the cargo box edge are screened out. Subsequently, the system calculates the shortest Euclidean distance between the bucket teeth and the cargo box. When the distance between the bucket teeth and the cargo box is less than the set safety distance threshold (0.3m), the strategy is adjusted, including raising the bucket height, adjusting the flip angle, and maintaining a safe distance from the cargo box. When the shovel loader reaches the target unloading position, the system controls the bucket flip angle so that the material can be unloaded into the cargo box smoothly. During the unloading process, the system first drives the flip cylinder mechanism to flip the bucket at the set angle to ensure that the material can be smoothly dumped into the cargo box. In order to improve the thoroughness of unloading and reduce material adhesion, the system has designed a bucket vibration control mechanism to control the bucket to vibrate at a specific frequency so that the adhered ore can be completely fallen off.

[0145] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0146] In an embodiment of the present invention, the pose of a scraper is generated by calculating the collected point cloud data of the roadway environment. The scraper is then path-controlled to reach the target location by combining the kinematic model of the articulated truck and the MPPI algorithm model. Based on the target location, environmental information data of the dump truck is collected, and features of the dump truck are extracted using a neural network. The extracted features are fused together. Based on the fused features of the dump truck, the pose of the dump truck is detected using the YOLOv7-tiny architecture, and the coordinates of the key points of the dump truck's cargo area are output. Based on the key point coordinates and the lidar point cloud data, the EPnP algorithm is used to calculate the 3D pose of the dump truck's cargo area. Based on the 3D pose, a dynamic unloading strategy is used to calculate key parameters describing the relative position of the scraper and the dump truck's cargo area. By combining key parameters and real-time environmental perception technology, the bucket posture of the scraper is dynamically adjusted to prevent the bucket from scratching the dump truck cargo box. According to the adjusted bucket posture, the global coordinates of the bucket teeth are calculated through the vehicle model. Based on the global coordinates, the bucket teeth of the scraper are adjusted to control the scraper to autonomously unload onto the dump truck. This achieves fully autonomous control, efficient path planning, accurate posture estimation and dynamic adjustment, reducing human intervention and improving work efficiency.

[0147] Reference Manual Figure 6 , which shows a structural schematic diagram of a control system for autonomous unloading of a scraper provided by the present invention.

[0148] The present invention further provides a control system 20 for autonomous unloading of a scraper, which is applied to the above-mentioned control method for autonomous unloading of a scraper, comprising:

[0149] Processor 201.

[0150] The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201 , the control method for autonomous unloading of a scraper loader according to the method embodiment is implemented.

[0151] The control system 20 for autonomous unloading of a scraper provided by the present invention can execute the above-mentioned control method for autonomous unloading of a scraper and achieve the same or similar technical effects. To avoid repetition, the present invention will not elaborate on them.

[0152] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0153] In an embodiment of the present invention, the pose of a scraper is generated by calculating the collected point cloud data of the roadway environment. The scraper is then path-controlled to reach the target location by combining the kinematic model of the articulated truck and the MPPI algorithm model. Based on the target location, environmental information data of the dump truck is collected, and features of the dump truck are extracted using a neural network. The extracted features are fused together. Based on the fused features of the dump truck, the pose of the dump truck is detected using the YOLOv7-tiny architecture, and the coordinates of the key points of the dump truck's cargo area are output. Based on the key point coordinates and the lidar point cloud data, the EPnP algorithm is used to calculate the 3D pose of the dump truck's cargo area. Based on the 3D pose, a dynamic unloading strategy is used to calculate key parameters describing the relative position of the scraper and the dump truck's cargo area. By combining key parameters and real-time environmental perception technology, the bucket posture of the scraper is dynamically adjusted to prevent the bucket from scratching the dump truck cargo box. According to the adjusted bucket posture, the global coordinates of the bucket teeth are calculated through the vehicle model. Based on the global coordinates, the bucket teeth of the scraper are adjusted to control the scraper to autonomously unload onto the dump truck. This achieves fully autonomous control, efficient path planning, accurate posture estimation and dynamic adjustment, reducing human intervention and improving work efficiency.

[0154] It should be understood that the processor in the embodiments of the present invention may be a central processing unit (CPU), and the processor may also be 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.

[0155] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0156] The above embodiments can be implemented in whole or in part through software, hardware (such as circuits), firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function according to the embodiments of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired method (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available media can be magnetic media (such as floppy disks, hard disks, tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0157] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0158] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0159] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0160] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0161] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0162] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, and can be electrical, mechanical, or other forms.

[0163] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0164] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0165] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and other media that can store program codes.

[0166] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the control method for autonomous unloading of a scraper loader as described in the method embodiment is implemented.

[0167] The computer-readable storage medium provided by the present invention can implement the steps and effects of the control method for autonomous unloading of a scraper loader in the above-mentioned method embodiment. To avoid repetition, the present invention will not elaborate on them.

[0168] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0169] In an embodiment of the present invention, the pose of a scraper is generated by calculating the collected point cloud data of the roadway environment. The scraper is then path-controlled to reach the target location by combining the kinematic model of the articulated truck and the MPPI algorithm model. Based on the target location, environmental information data of the dump truck is collected, and features of the dump truck are extracted using a neural network. The extracted features are fused together. Based on the fused features of the dump truck, the pose of the dump truck is detected using the YOLOv7-tiny architecture, and the coordinates of the key points of the dump truck's cargo area are output. Based on the key point coordinates and the lidar point cloud data, the EPnP algorithm is used to calculate the 3D pose of the dump truck's cargo area. Based on the 3D pose, a dynamic unloading strategy is used to calculate key parameters describing the relative position of the scraper and the dump truck's cargo area. By combining key parameters and real-time environmental perception technology, the bucket posture of the scraper is dynamically adjusted to prevent the bucket from scratching the dump truck cargo box. According to the adjusted bucket posture, the global coordinates of the bucket teeth are calculated through the vehicle model. Based on the global coordinates, the bucket teeth of the scraper are adjusted to control the scraper to autonomously unload onto the dump truck. This achieves fully autonomous control, efficient path planning, accurate posture estimation and dynamic adjustment, reducing human intervention and improving work efficiency.

[0170] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

[0171] There are a few points to note:

[0172] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention. Other structures may refer to conventional designs.

[0173] (2) For the sake of clarity, the thickness of layers or regions in the drawings used to describe the embodiments of the present invention are exaggerated or reduced, that is, these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region, or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or intervening elements may be present.

[0174] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to form new embodiments.

[0175] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for controlling the autonomous unloading of a scraper, characterized in that: include: S1: Collecting point cloud data of roadway environment; S2: Calculating the position of the scraper using a nine-axis inertial measurement unit based on the roadway environment point cloud data; S3: Based on the position of the scraper, combined with the kinematic model of the articulated truck and the MPPI algorithm model, the scraper is controlled to reach the target position; S4: According to the target position, environmental information data of the dump truck is collected, and features of the environmental information data are extracted using a neural network to determine features of the dump truck, wherein the environmental information data includes lidar point cloud data and RGB image data; S5: The extracted dump truck features are fused through the cross-feature fusion block to obtain the dump truck fusion features, and YOLOv7-tiny is used as the architecture to perform pose detection on the dump truck based on the dump truck fusion features, and output the key point coordinates of the dump truck cargo area; S6: Calculate the 3D pose of the dump truck container using the EPnP algorithm based on the key point coordinates and the lidar point cloud data; S7: Calculating key parameters describing the relative positions of the scraper and the dump truck container using a dynamic unloading strategy based on the 3D pose; S8: Combining the key parameters and real-time environment perception technology, dynamically adjusting the bucket posture of the scraper to prevent the bucket from scratching the dump truck container; S9: Calculate the global coordinates of the bucket teeth using the vehicle model according to the adjusted bucket posture; S10: adjusting the bucket teeth of the scraper according to the global coordinates to control the scraper to autonomously unload onto the dump truck.

2. The control method for autonomous unloading of a scraper according to claim 1, characterized in that: The calculation formula for calculating the position and posture of the scraper by the nine-axis inertial measurement unit in S2 is specifically: Among them, v k+1 represents the speed of the scraper at time k+1, v k represents the speed of the scraper at time k, a k represents the acceleration of the scraper at time k, Δt represents the control period, and θ k+1 represents the heading angle of the scraper at time k+1, θ k represents the heading angle of the scraper at time k, ω z represents the angular velocity of the scraper in the Z-axis (perpendicular to the ground), x k+1 represents the position of the scraper on the x-axis at time k+1, x k represents the position of the scraper on the x-axis at time k, and y k+1 represents the position of the scraper on the y-axis at time k+1, k represents the position of the scraper on the y-axis at time k.

3. The control method for autonomous unloading of a scraper according to claim 1, characterized in that: The S3 specifically includes: S301: Establishing the kinematic model of the articulated vehicle: in, represents the kinematic model of the articulated vehicle, represents the state matrix, represents the input matrix, u represents the input vector, v f represents the speed of the front vehicle, Φ f Indicates the heading angle of the front vehicle, v r represents the speed of the rear vehicle, θ f Indicates the heading angle difference between the front and rear vehicles, l f Indicates the axle length of the front vehicle body, l r represents the axis length of the rear body, X represents the state vector, x f Indicates the horizontal coordinate of the front vehicle body, y f represents the ordinate of the front vehicle body, T represents the transposition, a represents the acceleration of the front vehicle body, ω represents the angular velocity of the articulation angle, Φ r Indicates the heading angle of the rear vehicle body; S302: Combining the articulated vehicle kinematic model and the MPPI algorithm model, setting a cost function: c(x)=obstacle(x)+Speed(x)+Forward(x); r(t)=(x(t),y(t),z(t)); r 2D (t)=(x(t),y(t))=Proj xy (r(t)); Forward(x)=-w*v x ; Where c(x) represents the cost function, obstacle(x) represents the obstacle cost function, Speed(x) represents the speed cost function, Forward(x) represents the forward direction cost function, ξ1 and ξ2 represent weighting factors, exp represents the exponential function, mindist represents the minimum distance between the vehicle and the obstacle, hit represents a binary variable, and distance represents the difference between the vehicle coordinates (x, y) and the obstacle coordinates (obs x ,obs y ), r(t) represents the position vector of the vehicle in three-dimensional space at time t, x(t), y(t) and z(t) represent the position coordinates of the vehicle on the x, y and z axes at time t respectively, r 2D (t) represents the position vector of the vehicle on the two-dimensional plane at time t, Proj xy (r(t)) represents the projection of the three-dimensional position r(t) onto the xy plane, v 2D (t) represents the speed of the vehicle on the two-dimensional plane at time t, and Represent the speed components of the vehicle in the x and y directions respectively, Speed(t) represents the total speed of the vehicle at time t, λ represents the weighting factor, represents the rate of change of the vehicle's speed on the two-dimensional plane, that is, acceleration, and ω represents the weighting factor; S303: With the goal of minimizing the cost function value, the scraper is controlled to reach the target position.

4. The control method for autonomous unloading of a scraper according to claim 1, characterized in that: After S4, the method further includes: Preprocessing the environmental information data, the preprocessing specifically includes: Combine the rotation matrix and translation matrix to transform the lidar coordinates to the camera coordinates: Among them, [X C ,Y C ,Z C ] represents the coordinates in the camera coordinate system, R represents the rotation matrix, T represents the translation matrix, [X L ,Y L ,Z L ] represents the coordinates in the laser radar coordinate system; Project the transformed lidar coordinates into the camera's 2D image coordinate system: Among them, (x, y) represents the two-dimensional coordinates in the two-dimensional image coordinate system, f x Indicates the focal length of the camera on the x-axis, f y Indicates the focal length of the camera on the y-axis, c x Indicates the coordinate of the camera's optical center on the x-axis, c y Indicates the coordinate of the camera's optical center on the y-axis; The laser radar point cloud data is converted into a depth image having the same size as the RGB image data according to the two-dimensional image coordinate system.

5. The control method for autonomous unloading of a scraper according to claim 4, characterized in that: The S4 is specifically: The dump truck features are extracted from the RGB image data and the depth image through a neural network, wherein the neural network includes two independent convolutional neural network branches with shared parameters, and the convolutional neural network branches include: an RGB branch and a depth image branch, the RGB branch is responsible for extracting the texture features of the dump truck, and the depth image branch is responsible for extracting the geometric features of the dump truck.

6. The control method for autonomous unloading of a scraper according to claim 4, characterized in that: The calculation formula of the dump truck fusion feature in S5 is specifically: F fusion =W1·F RGB +W2·F Depth ; Among them, F fusion represents the dump truck fusion feature, W1 and W2 both represent adaptive weight factors, F RGB represents the features extracted from RGB images, F Depth Represents features extracted from depth images.

7. The control method for autonomous unloading of a scraper according to claim 1, characterized in that: After S5, the following steps are further included: By introducing a Transformer-based self-attention mechanism, the focus on the dump truck cargo area is enhanced; The calculation formula of the Transformer's self-attention mechanism is specifically as follows: Among them, A represents the attention weight matrix, softmax represents the normalization function, Q represents the query matrix, K represents the key matrix, V represents the value matrix, and d k represents the dimensions of the query and key vectors, and T represents the transpose.

8. The control method for autonomous unloading of a scraper according to claim 1, characterized in that: After S6, the method further includes: The 3D posture is smoothed by Kalman filtering to reduce the impact of noise on the accuracy of posture detection.

9. The control method for autonomous unloading of a scraper according to claim 1, characterized in that: The calculation formula of the global coordinates of the bucket teeth in S9 is specifically: in, represents the global coordinates of the bucket teeth, Indicates the center position of the scraper, R veh represents the vehicle posture rotation matrix, Indicates the position of the scraper arm rotation point, R arm represents the rotation matrix of the lifting angle of the scraper arm, R bucket represents the scraper bucket flip angle matrix, Indicates the fixed position of the scraper bucket teeth in the bucket coordinate system.

10. A control system for autonomous unloading of a scraper, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the control method for autonomous unloading of a scraper according to any one of claims 1 to 9 is implemented.

Images