A method, device, equipment and medium for laterally grabbing a box based on a dual-arm robot

Abstract

The application provides a kind of based on dual-arm robot lateral grabbing box method, box is equipped with two-dimensional code, in humanoid dual-arm robot is grabbed box by mechanical arm, comprising the following steps: box is photographed by camera, and the shooting image and point cloud image of box are obtained;The pose of two-dimensional code in the shooting image is obtained;The posture of mechanical arm grabbing box is calculated according to the pose of two-dimensional code;The coordinates of the center point of two-dimensional code in point cloud image under camera coordinate system are determined;The grabbing point of box is calculated according to the coordinates of the center point of two-dimensional code in point cloud image under camera coordinate system;The posture of mechanical arm grabbing box and the grabbing point of box are used to control mechanical arm to grab box, the application can realize the quick and accurate detection to the tilt pose of box, and automatically adjust the end of mechanical arm grabbing posture according to the tilt condition of box, greatly improve the success rate of grabbing.

Description

Technical Field

[0001] This invention relates to the field of robot control, and in particular to a method, apparatus, device, and medium for lateral grasping of a box using a dual-arm robot. Background Technology

[0002] Humanoid robots are still under development. With the continuous improvement of their basic walking and balance abilities, they can now perform some simple tasks, such as picking up and carrying objects. Equipped with vision systems and object recognition algorithms, they can autonomously navigate environments such as warehouses, transporting goods to designated locations, thereby reducing the burden on human labor and improving work efficiency.

[0003] When a robot is handling object handling tasks, it needs to obtain the pose of the object relative to the robot body through the perception module, and calculate the gripping point and gripping pose of the robot's gripper end based on the actual size of the object.

[0004] Currently, there are two common approaches to grasping in humanoid dual-arm robots. The traditional approach typically involves manually creating a 3D model of the object to be grasped and building a grasping database based on the grasping process. During actual grasping, the object model needs to be detected in real time and retrieved from the grasping database to obtain the grasping posture. This method cannot grasp objects outside the database, lacks basic robustness, and because the robot can only detect objects from one angle during use, the accuracy of retrieving the object from the database is low.

[0005] Deep learning-based grasping methods, such as the patent with patent number CN 111347411 A, entitled "Deep Learning-Based Three-Dimensional Vision Recognition and Grasping Method for Dual-Arm Collaborative Robots," have the following drawbacks: low accuracy in recognizing object poses, limited range of recognizable object poses, and a high error rate in calculating grasping poses through deep learning. Furthermore, they cannot ensure the determinism of the overall grasping process and are unsuitable for industrial scenarios. Summary of the Invention

[0006] The purpose of this invention is to provide a method, apparatus, device, and medium for lateral grasping of boxes based on a dual-arm robot, thereby improving the accuracy of box grasping.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for lateral grasping of a box using a dual-arm robot, wherein the box is provided with a QR code, and the humanoid dual-arm robot grasps the box using its robotic arm, comprising the following steps: taking a picture of the box with a camera to obtain a captured image and a point cloud image of the box; obtaining the pose of the QR code in the captured image; calculating the posture of the robotic arm grasping the box based on the pose of the QR code; determining the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system; calculating the grasping point of the box based on the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system; and controlling the robotic arm to grasp the box based on the posture of the robotic arm grasping the box and the grasping point of the box.

[0008] Furthermore, before taking pictures of the box with the camera, the following steps are also included: performing hand-eye calibration on the robotic arm and the camera; establishing a camera coordinate system based on the camera and a robot coordinate system based on the humanoid dual-arm robot; and confirming the transformation relationship between the camera coordinate system and the robot coordinate system.

[0009] Furthermore, obtaining the pose of the QR code in the captured image includes: establishing an image recognition function; and recognizing the coordinates and pose of the QR code in the camera coordinate system based on the image.

[0010] Further, determining the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system includes: traversing and checking each coordinate of the point cloud image, removing zero vectors and non-finite points in the point cloud image; removing outliers in the point cloud image; performing voxel downsampling on the point cloud image; calculating the normal vector of each point in the point cloud image after voxel downsampling; and calculating the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system based on the pixel coordinates of the center point of the QR code in the captured image.

[0011] Further, the step of calculating the grab point of the box based on the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system includes: setting the number of neighboring points n to be found; based on the K-nearest neighbor search method, finding the n point clouds in the point cloud image that are closest to the center point of the QR code, and taking the found point clouds as the neighboring points; calculating the dot product of the normal of each neighboring point and the center point of the QR code, and obtaining the neighboring point with the largest angle difference from the center point of the QR code; calculating the constant term D of the plane equation based on the normal vector of the neighboring point with the largest angle difference and the coordinates of the center point of the QR code in the camera coordinate system; calculating the plane containing the side of the box with the QR code in the point cloud image based on the constant term D of the plane equation and the normal vector of the neighboring point with the largest angle difference; traversing all point clouds of the plane containing the side of the box with the QR code, and calculating the grab point of the box with the QR code based on the coordinates of the center point of the QR code in the point cloud image. Using the distance from the point cloud to the plane and the angle between the point cloud and the plane's normal, point clouds close to the side of the box with the QR code are filtered out. A maximum neighborhood distance and a minimum number of point clouds are set, and the filtered point clouds are clustered based on these parameters. A second filtering is performed based on the number of point clouds in each cluster and their distance from the center point, thereby segmenting the point cloud image corresponding to the side of the box with the QR code. Based on the point clouds corresponding to the side of the box with the QR code, four edges of the box's upper surface are fitted, and the coordinates of the four corner points of the box's upper surface are obtained through these four edges. The coordinates of the center point of each edge of the box's upper surface are calculated based on the corner point coordinates. The center points of two opposite edges on the box's upper surface are selected as gripping calibration points. Points at a set distance from the gripping calibration points are set along the direction opposite to the normal direction of the QR code's center point.

[0012] Further, the step of calculating the posture of the robotic arm grasping the box based on the pose of the QR code includes: dragging the robotic arm to teach the grasping posture, obtaining the grasping posture of the end effector of the robotic arm in the robot coordinate system; calculating the relationship between the grasping posture of the end effector of the robotic arm and the QR code based on the transformation relationship between the camera coordinate system and the robot coordinate system and the grasping posture of the end effector of the robotic arm in the base coordinate system; and calculating the posture of the robotic arm grasping the box based on the relationship between the grasping posture of the end effector of the robotic arm and the QR code.

[0013] Furthermore, the number of neighboring points n is 20.

[0014] On the other hand, a device for lateral grasping of a box based on a dual-arm robot is provided, comprising: a camera unit for capturing images of the box and obtaining captured images and point cloud images of the box; a captured image judgment unit for obtaining the pose of the QR code in the captured image; a first calculation unit for calculating the posture of the robotic arm grasping the box based on the pose of the QR code; a point cloud image judgment unit for determining the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system; a second calculation unit for calculating the grasping point of the box based on the coordinates of the center point of the QR code in the camera coordinate system; and a robotic arm control unit for controlling the robotic arm to grasp the box based on the posture of the robotic arm grasping the box and the grasping point of the box.

[0015] On the other hand, an electronic device is provided, comprising: a processor and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the above-described box-grabbing method.

[0016] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction, at least one program, code set, or instruction set is stored therein, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement the above-described box-grabbing method.

[0017] Analysis shows that the present invention provides a method, device, equipment and medium for lateral grasping of a box based on a dual-arm robot, which can quickly and accurately detect the tilt position of the box and automatically adjust the grasping posture of the end of the robotic arm according to the tilt of the box, greatly improving the grasping success rate. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. Wherein:

[0019] Figure 1 A flowchart of an embodiment of the present invention. Detailed Implementation

[0020] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. Various examples are provided by way of explanation and not by way of limitation. Indeed, those skilled in the art will recognize that modifications and variations can be made to the invention without departing from its scope or spirit. For example, a feature shown or described as part of one embodiment may be used in another embodiment to produce yet another embodiment. Therefore, it is desirable that the invention encompass such modifications and variations falling within the scope of the appended claims and their equivalents.

[0021] The accompanying drawings illustrate one or more examples of the invention. The detailed description uses numerals and letters to refer to features in the drawings. Similar or analogous reference numerals in the drawings and description have been used to refer to similar or analogous parts of the invention. As used herein, the terms “first,” “second,” “third,” and “fourth,” etc., are used interchangeably to distinguish one component from another and are not intended to indicate the location or importance of a single component.

[0022] like Figure 1 As shown, according to an embodiment of the present invention, a method for lateral grasping of a box based on a dual-arm robot is provided. The box has a QR code. A humanoid dual-arm robot grasps the box using its robotic arm. Hand-eye calibration is performed on the robotic arm and camera to ensure the camera is located outside the robotic arm. A camera coordinate system and a robot coordinate system based on the camera are established. The transformation relationship between the camera coordinate system and the robot coordinate system is confirmed, denoted as...

[0023] The aforementioned dual-armed robots can include humanoid robots or collaborative robots.

[0024] The QR code mentioned above can be an Aruco code, an ARTag code, or an AprilTag code.

[0025] Includes the following steps:

[0026] Step S101: Take a picture of the box with a camera to obtain a picture image and a point cloud image of the box;

[0027] The images mentioned above are images produced by the camera, and RGB images are generally used as the captured images.

[0028] The specific steps for photographing the container are as follows: the camera must be positioned so that the QR code on the container is visible during the photograph, and the container must also be within the robot's arm's range of motion. In industrial processes, it is usually necessary for preceding processes to ensure that the QR code is facing upwards. If this cannot be guaranteed, a QR code must be affixed to three sides of one corner of the container to ensure that the camera can capture the QR code.

[0029] Step S102: Obtain the pose of the QR code in the captured image;

[0030] Specifically, an image recognition function is established; based on the image recognition function, the coordinates and pose of the QR code in the camera coordinate system are recognized.

[0031] The calculation is performed using the `solvePnP` function interface from the OpenCV library. The operator used is the IPPE (Induced Depth Point with a Pinhole Camera) iterative operator, suitable for scenes containing square markers in the image. This calculates the coordinates of the QR code's center point in the camera coordinate system, as well as the QR code's pose in the camera coordinate system, denoted as […].

[0032] Step S103: Calculate the posture of the robotic arm to grasp the box based on the pose of the QR code;

[0033] The posture of the robotic arm in grasping the box is calculated based on the pose of the QR code, including:

[0034] The robotic arm is dragged to teach the grasping posture, thus obtaining the grasping posture of the robotic arm's end effector in the robot coordinate system.

[0035] Based on the transformation relationship between the camera coordinate system and the robot coordinate system, and the grasping posture of the robotic arm's end effector in the base coordinate system, the relationship between the grasping posture of the robotic arm's end effector and the QR code is calculated. The calculation method is as follows: Will The value is stored as a constant. The gripping pose of the robotic arm for each grabbing of the box is calculated using the following formula:

[0036] The above This describes the attitude relationship between the robotic arm's end effector's grasping posture and the QR code. This relationship is obtained through manual teaching and saved for subsequent calculations of the robotic arm's grasping pose. The robotic arm's grasping pose can then be obtained from the QR code's posture, thus simplifying the calculation process.

[0037] The gripping posture of the robotic arm is calculated based on the relationship between the gripping posture of the robotic arm's end effector and the QR code.

[0038] Step S104: Determine the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system;

[0039] Determining the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system includes: traversing and checking each coordinate of the point cloud image, and removing zero vectors and non-finite points (such as infinity or nan values) in the point cloud image.

[0040] Remove outliers from point cloud images; typically, points whose distance to the 50 nearest neighbors is greater than 0.05 are considered outliers and are removed accordingly.

[0041] Voxel downsampling is performed on the point cloud image to reduce the number of points. The normal vector of each point in the voxel-downsampled point cloud image is then calculated. Specifically, a K-nearest neighbor search is used to search for neighboring points, typically considering the 100 nearest neighbors for each point. Two orthogonal vectors are calculated based on these neighboring points, forming a tangent plane. The product of these two tangent plane vectors is then calculated to obtain the normal vector of each point in the point cloud.

[0042] The coordinates of the center point of the QR code in the point cloud image in the camera coordinate system are calculated based on the pixel coordinates of the center point of the QR code in the captured image.

[0043] The pixel coordinates mentioned above refer to the pixel coordinates of the center point of the QR code on the captured image. Based on the pixel coordinates and the intrinsic parameter matrix, the pixel coordinates can be converted into coordinates in the corresponding camera coordinate system.

[0044] In practical operation, the RealSense depth camera from Inter can be used. It provides an interface to align the camera image and point cloud, bind the image pixel information and depth information, correct the camera distortion through distortion coefficients, and calculate the conversion between pixel coordinates and camera coordinates based on the focal length and principal point in the camera intrinsic parameter matrix.

[0045] Step S105: Calculate the grab point of the box based on the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system;

[0046] The calculation of the grab points of the box based on the point cloud coordinates of the QR code includes:

[0047] Set the number of nearest neighbors n to be found; typically, the number of nearest neighbors n is 20.

[0048] Based on the K-nearest neighbor search method, find the n point clouds that are closest to the center point of the QR code in the image, and use the found point clouds as the nearest points;

[0049] Calculate the dot product of the normals of each neighboring point and the center point of the QR code to obtain the neighboring point with the largest angular difference from the center point of the QR code.

[0050] Based on the normal vector of the nearest point with the largest angular difference and the coordinates of the QR code center point in the camera coordinate system, the constant term D of the plane equation is calculated.

[0051] Based on the constant term D of the plane equation and the normal vector of the nearest point with the largest angle difference, the plane containing the side of the box with the QR code in the point cloud image is calculated.

[0052] Regarding the plane equation, it is used to define a plane in three-dimensional space: ax + by + cz + D = 0, where (a, b, c) are the plane normal vectors, and D is a constant term. In three-dimensional space, any plane can be determined by a normal vector pointing to the plane and a point on the plane. The normal vector is a vector perpendicular to the plane, and the constant term D can be calculated by taking the dot product of the normal vector and the position vector of any point on the plane. Therefore, based on the normal vectors of the nearest points with the largest angular differences and the point cloud coordinates of the center point of the QR code, the constant term D of the plane equation can be calculated. Once the normal vector and the constant term D are determined, a plane can be completely defined.

[0053] Traverse all point clouds on the plane containing the side of the box with the QR code, and filter out the point clouds that are close to the side of the box with the QR code based on the distance of all point clouds on the plane containing the side of the box with the QR code to the plane and the angle between the point clouds and the plane normal.

[0054] The specific requirements for proximity in the point cloud filtering process for the side of the box with the QR code can be set according to the actual situation. For example, if the distance from the point to the plane is less than 0.005 or the angle difference is less than 30 degrees, then the point is considered to be on the plane, which is the point cloud of the side of the box with the QR code.

[0055] Set the maximum neighborhood distance and the minimum number of point clouds, and cluster the filtered point clouds based on the maximum neighborhood distance and the minimum number of point clouds; the DBSCAN algorithm is usually used for clustering.

[0056] Secondary filtering is performed based on the number of point clouds in each category after clustering and the distance from the center point, thereby segmenting the point cloud image corresponding to the side of the box with the QR code attached.

[0057] Based on the point cloud corresponding to the side of the box with the QR code attached, the four sides of the upper surface of the box are fitted, and the coordinates of the four corner points of the upper surface of the box are obtained through the four sides of the upper plane of the box.

[0058] The coordinates of the center point of each side of the upper surface of the box are calculated based on the corner coordinates;

[0059] Along the direction opposite to the normal direction of the center point of the QR code, a point at a set distance from the grasping calibration point is set as the grasping point;

[0060] Taking the case where the side of the box with the QR code is parallel to the x and y axes of the base coordinate system as an example, the center points of two opposite sides on the upper surface of the box are selected as the gripping calibration points. The x and y coordinates of the gripping calibration points are obtained by converting point cloud coordinates. Since the x and y coordinates of the gripping calibration points are the same as those of the gripping points, the x and y coordinates of the gripping points can be obtained by acquiring the x and y coordinates of the gripping calibration points. The method for calculating the z-axis coordinates of the gripping points is as follows: move a certain distance in the opposite direction of the normal vector of the box's point cloud plane. This distance depends on the thickness of the gripping box and the thickness of the grippers, i.e., set a distance. Transform the coordinates of the point after moving a certain distance to the base coordinate system to obtain the z-axis coordinates of the gripping points in the base coordinate system.

[0061] Step S106: Control the robotic arm to grasp the box according to the posture of the robotic arm grasping the box and the grasping point of the box.

[0062] This invention also discloses a device for lateral grasping of a box based on a dual-arm robot, comprising: a camera unit for capturing images of the box and obtaining captured images and point cloud images of the box; a captured image judgment unit for obtaining the pose of a QR code in the captured image; a first calculation unit for calculating the posture of the robotic arm grasping the box based on the pose of the QR code; a point cloud image judgment unit for determining the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system; a second calculation unit for calculating the grasping point of the box based on the coordinates of the center point of the QR code in the camera coordinate system; and a robotic arm control unit for controlling the robotic arm to grasp the box based on the posture of the robotic arm grasping the box and the grasping point of the box.

[0063] The present invention also discloses an electronic device comprising: a processor and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the above-described grasping box method.

[0064] The present invention also discloses a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement the above-described box-grabbing method.

[0065] As can be seen from the above description, the embodiments of the present invention achieve the following technical effects: The present invention is mainly aimed at the use of humanoid dual-arm robots in industrial scenarios, where high accuracy in detection and grasping precision are required. The present invention uses the most reliable QR code detection algorithm to perform pose localization of the box and introduces point cloud plane detection, so that it can be applied to the calculation of grasping points for boxes of any size without changing parameters.

[0066] This invention addresses the grasping task of a dual-arm robot with a movable chassis. In industrial scenarios, the container itself is often tilted, and the movement of the robot chassis exacerbates the tilt between the container and the robotic arm's coordinate system. To address this, this invention optimizes the grasping strategy for container tilt in grasping scenarios. This optimization enables rapid and accurate detection of the container's tilt posture and automatically adjusts the robotic arm's end effector's grasping posture based on the container's tilt, significantly improving the grasping success rate.

[0067] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for lateral grasping of a box using a dual-arm robot, wherein the box is equipped with a QR code, and a humanoid dual-arm robot grasps the box using its robotic arms, characterized in that... Includes the following steps: The box is photographed with a camera to obtain photographic images and point cloud images of the box; Obtain the pose of the QR code in the captured image; The posture of the robotic arm grasping the box is calculated based on the pose of the QR code; Determine the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system; The grab point of the box is calculated based on the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system. The robotic arm is controlled to grasp the box based on its posture and the grasping point of the box; The step of calculating the grab point of the box based on the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system includes: Set the number of nearest neighbors n to be found; Based on the K-nearest neighbor search method, find the n point clouds in the point cloud image that are closest to the center point of the QR code, and use the found point clouds as the nearest points; Calculate the dot product of the normal to the center point of the QR code for each of the neighboring points, and obtain the neighboring point with the largest angular difference from the center point of the QR code. Based on the normal vector of the nearest point with the largest angle difference and the coordinates of the center point of the QR code in the camera coordinate system, the constant term D of the plane equation is calculated. Based on the constant term D of the plane equation and the normal vector of the nearest point with the largest angle difference, the plane containing the side of the box with the QR code in the point cloud image is calculated. Traverse all point clouds on the plane containing the side of the box with the QR code, and filter out the point clouds that are close to the side of the box with the QR code based on the distance of all point clouds on the plane containing the side of the box with the QR code to the plane and the angle between the point clouds and the plane normal. Set the maximum neighborhood distance and the minimum number of point clouds, and cluster the filtered point clouds based on the maximum neighborhood distance and the minimum number of point clouds; Secondary filtering is performed based on the number of point clouds in each category after clustering and the distance from the center point, thereby segmenting the point cloud image corresponding to the side of the box with the QR code attached. Based on the point cloud corresponding to the side of the box with the QR code attached, the four sides of the upper surface of the box are fitted, and the coordinates of the four corner points of the upper surface of the box are obtained through the four sides of the upper plane of the box. The coordinates of the center point of each side of the upper surface of the box are calculated based on the corner coordinates; The center points of two opposite sides on the upper surface of the box are selected as the gripping calibration points; Along the direction opposite to the normal direction of the center point of the QR code, a point at a set distance from the grasping calibration point is set as the grasping point.

2. The method for lateral grasping of a box based on a dual-arm robot according to claim 1, characterized in that, Before photographing the enclosure with a camera, the following steps are also included: Hand-eye calibration is performed on the robotic arm and the camera; Establish a camera coordinate system based on the camera and a robot coordinate system based on the humanoid dual-arm robot; Confirm the transformation relationship between the camera coordinate system and the robot coordinate system.

3. The method for lateral grasping of a box based on a dual-arm robot according to claim 2, characterized in that, The step of obtaining the pose of the QR code in the captured image includes: Establish an image recognition function; The image recognition function identifies the coordinates and pose of the QR code in the camera coordinate system.

4. The method for lateral grasping of a box based on a dual-arm robot according to claim 3, characterized in that, Determining the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system includes: Iterate through and examine each coordinate of the point cloud image, and remove zero vectors and non-finite points from the point cloud image; Remove outliers from the point cloud image; Voxel downsampling is performed on the point cloud image; Calculate the normal vector of each point in the point cloud image after voxel downsampling; The coordinates of the center point of the QR code in the point cloud image in the camera coordinate system are calculated based on the pixel coordinates of the center point of the QR code in the captured image.

5. A method for lateral grasping of a box based on a dual-arm robot according to claim 1, characterized in that, The step of calculating the posture of the robotic arm grasping the box based on the pose of the QR code includes: The robotic arm is dragged to teach the grasping posture, thereby obtaining the grasping posture of the end effector of the robotic arm in the robot coordinate system; Based on the transformation relationship between the camera coordinate system and the robot coordinate system and the grasping posture of the end effector of the robotic arm in the base coordinate system, the relationship between the grasping posture of the end effector of the robotic arm and the QR code is calculated. The gripping posture of the robotic arm is calculated based on the relationship between the gripping posture of the robotic arm's end effector and the QR code.

6. A method for lateral grasping of a box based on a dual-arm robot according to claim 1, characterized in that, The number of neighboring points, n, is 20.

7. A device for lateral grasping of a box using a dual-arm robot, applying the lateral grasping method for a box based on any one of claims 1-6, characterized in that, include: The camera unit captures images of the enclosure and obtains point cloud images of the enclosure using a camera. The image capture judgment unit obtains the pose of the QR code in the captured image; The first calculation unit calculates the posture of the robotic arm grasping the box based on the pose of the QR code; A point cloud image determination unit determines the coordinates of the center point of the QR code in the point cloud image in the camera coordinate system. The second calculation unit calculates the grab point of the box based on the coordinates of the center point of the QR code in the camera coordinate system; The robotic arm control unit controls the robotic arm to grasp the box based on the posture of the robotic arm grasping the box and the grasping point of the box.

8. An electronic device, characterized in that, The electronic device includes: a processor and a memory for storing executable instructions of the processor; The processor is configured to perform the box-grabbing method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction, which is loaded and executed by a processor to implement the gripping box method according to any one of claims 1-6.

Images