Joint mechanical arm calibration method and device and automatic calibration system

Abstract

The embodiment of the invention provides a joint mechanical arm calibration method and device and an automatic calibration system, and relates to the technical field of robot equipment.The joint mechanical arm drives a calibration assembly to move to different calibration points, and point cloud data corresponding to the calibration points are obtained through radar at the calibration points; determining a rough position of the calibration point according to a rough coordinate conversion relation between a mechanical arm coordinate system and a radar coordinate system, then screening out point cloud data belonging to an area where the calibration assembly is located, and fitting the screened point cloud data to obtain the position of a feature point in the calibration assembly; according to the obtained position and the relative position relation between the calibration point and the feature point, the fine position of the calibration point in the radar coordinate system is obtained, and finally the fine coordinate conversion relation between the mechanical arm coordinate system and the radar coordinate system of the radar is determined. By applying the method provided by the embodiment of the invention, the coordinate conversion relation between the radar coordinate system and the mechanical arm coordinate system can be accurately determined.

Description

Technical Field

[0001] This invention relates to the field of robotic equipment technology, and in particular to a method, apparatus and automatic calibration system for calibrating articulated robotic arms. Background Technology

[0002] Loading and unloading robots equipped with articulated robotic arms can grasp and stack goods in spatial settings such as truck beds and warehouses. In related technologies, loading and unloading robots often use radar to scan the surrounding environment to identify the location of the goods or the location where they need to be stacked. When controlling the movement of the articulated robotic arm, the loading and unloading robot controls the grasping component at the end of the articulated robotic arm to move to the designated position in the robot arm's coordinate system.

[0003] Taking the case of grasping goods as an example, the loading and unloading robot identifies the goods as being in the first position through radar. Since the first position is located in the radar coordinate system, if the loading and unloading robot directly controls the grasping component to move to the first position in the robotic arm coordinate system, it will be unable to grasp the goods. It is necessary to first determine the projection position of the first position in the radar coordinate system onto the robotic arm coordinate system based on the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system, and then control the grasping component to move to the projection position in the robotic arm coordinate system in order for the grasping component to successfully grasp the goods.

[0004] It is evident that accurately determining the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system is a prerequisite for the loading and unloading robot to effectively grasp and stack goods. Therefore, accurately determining the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system has become an urgent technical problem to be solved. Summary of the Invention

[0005] The purpose of this invention is to provide a method, apparatus, and automatic calibration system for calibrating articulated robotic arms, thereby improving the accuracy of loading and unloading robots in grasping and stacking goods. The specific technical solution is as follows:

[0006] In a first aspect, embodiments of this application provide a method for calibrating an articulated robotic arm, the method comprising:

[0007] For multiple different calibration points in the coordinate system of the articulated robotic arm, point cloud data obtained by radar scanning calibration components when the end effector of the articulated robotic arm is located at the calibration points are obtained, and these are used as the original point cloud data corresponding to the calibration points. The calibration components are set at the end effector.

[0008] For each calibration point, the approximate position of the calibration point in the radar coordinate system is determined based on the coarse coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system.

[0009] For each calibration point, based on the approximate location of the calibration point and the size of the calibration component, the point cloud data belonging to the region where the calibration component is located is filtered out from the original point cloud data corresponding to the calibration point, and used as the filtered point cloud data corresponding to the calibration point.

[0010] For each calibration point, the filtered point cloud data corresponding to the calibration point is fitted according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system. Based on the obtained position and the relative positional relationship between the calibration point and the feature point, the fine position of the calibration point in the radar coordinate system is obtained.

[0011] Based on the precise positions of each calibration point and their positions in the robotic arm coordinate system, a precise coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system is determined.

[0012] In one possible implementation, the step of filtering point cloud data belonging to the region where the calibration component is located from the original point cloud data corresponding to the calibration point, based on the approximate location of the calibration point and the size of the calibration component, as the filtered point cloud data corresponding to the calibration point, includes:

[0013] Based on the dimensions of the calibration component, a surrounding box capable of accommodating the calibration component is provided at the approximate location;

[0014] Point cloud data located within the bounding box are selected from the original point cloud data corresponding to the calibration point and used as the filtered point cloud data corresponding to the calibration point.

[0015] In one possible implementation, setting a surrounding box capable of accommodating the calibration component at the coarse location, based on the size of the calibration component, includes:

[0016] Based on the maximum spatial span of the calibrated components, a cube with a side length greater than the maximum spatial span is set as a bounding box at the approximate location.

[0017] In one possible implementation, the calibration component has a target end face and a fixed end face. The fixed end face of the calibration component is fixed to the flange at the end. The target end face is the end face of the calibration component that is away from the flange. The normal of the target end face of the calibration component is collinear with the normal of the flange. The feature point is the center of the target end face of the calibration component. The size of the target end face is larger than the size of the flange.

[0018] The step of fitting the filtered point cloud data corresponding to the calibration point according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system, and obtaining the fine position of the calibration point in the radar coordinate system based on the obtained position and the relative positional relationship between the calibration point and the feature point, includes:

[0019] The filtered point cloud data corresponding to the calibration points are fitted according to the shape of the target end face to obtain the fitted graphic of the target end face;

[0020] Based on the fitted graph, determine the position and normal of the center of the target end face in the radar coordinate system;

[0021] Based on the thickness of the calibration component, the obtained position, and the normal, the fine position of the calibration point in the radar coordinate system is calculated, wherein the thickness is the spatial span of the calibration component along the central axis.

[0022] In one possible implementation, the step of fitting the filtered point cloud data corresponding to the calibration points according to the shape of the target end face to obtain the fitted graphic of the target end face includes:

[0023] Determine the convex hull vertices in the filtered point cloud;

[0024] Based on each of the convex hull vertices, a fitted graph that passes through each of the convex hull vertices and has the same shape as the target end face is obtained, and this fitted graph is used as the fitted graph of the target end face.

[0025] In one possible implementation, the calibration component is a disk.

[0026] Secondly, embodiments of this application provide an automatic calibration system, the system including an articulated robotic arm, a radar, and a processor, wherein the end of the articulated robotic arm is provided with a calibration component;

[0027] The processor is configured to control the movement of the articulated robotic arm so that the end effector sequentially moves to multiple different calibration points in the robotic arm coordinate system of the articulated robotic arm; and whenever the end effector moves to a new calibration point, it sends a scan request to the radar, wherein each calibration point is located within the field of view of the radar;

[0028] The radar is configured to perform a scan in response to the scan request, obtain point cloud data, and send the point cloud data to the processor;

[0029] The processor is further configured to: acquire point cloud data obtained by scanning the calibration component with the radar when the end effector of the articulated robotic arm is at the calibration point, for each calibration point, as the original point cloud data corresponding to the calibration point; determine the approximate position of the calibration point in the radar coordinate system based on the approximate coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system for each calibration point; and filter out data belonging to the calibration component from the original point cloud data corresponding to the calibration point based on the approximate position of the calibration point and the size of the calibration component for each calibration point. The point cloud data of the specified area is used as the filtered point cloud data corresponding to the calibration point. For each calibration point, the filtered point cloud data corresponding to the calibration point is fitted according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system. Based on the obtained position and the relative positional relationship between the calibration point and the feature point, the fine position of the calibration point in the radar coordinate system is obtained. Based on the fine position of each calibration point and the position of each calibration point in the robotic arm coordinate system, the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system is determined.

[0030] Thirdly, embodiments of this application provide a calibration device for an articulated robotic arm, the device comprising:

[0031] The data acquisition module is used to acquire point cloud data obtained by a radar scanning calibration component when the end of the articulated robotic arm is located at multiple different calibration points in the coordinate system of the articulated robotic arm, and use the calibration component as the original point cloud data corresponding to the calibration point. The calibration component is set at the end.

[0032] The coarse position determination module is used to determine the coarse position of each calibration point in the radar coordinate system based on the coarse coordinate transformation relationship between the robot arm coordinate system and the radar coordinate system.

[0033] The data filtering module is used to filter out point cloud data belonging to the region where the calibration component is located from the original point cloud data corresponding to each calibration point, based on the approximate location of the calibration point and the size of the calibration component, and use it as the filtered point cloud data corresponding to the calibration point.

[0034] The fine position determination module is used to fit the filtered point cloud data corresponding to each calibration point according to the shape of the calibration component, to obtain the position of the feature point in the calibration component in the radar coordinate system, and to obtain the fine position of the calibration point in the radar coordinate system based on the obtained position and the relative positional relationship between the calibration point and the feature point.

[0035] The relationship determination module is used to determine the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system of the radar based on the fine position of each calibration point and the position of each calibration point in the robotic arm coordinate system.

[0036] In one possible implementation, the data filtering module includes:

[0037] A bounding box setting submodule is used to set a bounding box capable of accommodating the calibration component at the coarse location, based on the size of the calibration component;

[0038] The data filtering submodule is used to filter out the point cloud data located within the bounding box from the original point cloud data corresponding to the calibration point, and use it as the filtered point cloud data corresponding to the calibration point.

[0039] In one possible implementation, the bounding box setting submodule is specifically used for:

[0040] Based on the maximum spatial span of the calibration component, a cube with a side length greater than the maximum spatial span is set as a bounding box at the approximate location;

[0041] The calibration component has a target end face and a fixed end face. The fixed end face of the calibration component is fixed to the flange at the end. The target end face is the end face of the calibration component that is away from the flange. The normal of the target end face of the calibration component is collinear with the normal of the flange. The feature point is the center of the target end face of the calibration component. The size of the target end face is larger than the size of the flange.

[0042] In one possible implementation, the fine position determination module includes:

[0043] The data fitting submodule is used to fit the filtered point cloud data corresponding to the calibration point according to the shape of the target end face to obtain the fitted graphic of the target end face.

[0044] The normal direction determination submodule is used to determine the position and normal of the center of the target end face in the radar coordinate system based on the fitted graph.

[0045] The position calculation submodule is used to calculate the fine position of the calibration point in the radar coordinate system based on the thickness of the calibration component, the obtained position and normal, wherein the thickness is the spatial span of the calibration component along the central axis direction;

[0046] In one possible implementation, the data fitting submodule is specifically used for:

[0047] Determine the convex hull vertices in the filtered point cloud;

[0048] Based on each of the convex hull vertices, a fitted graph that passes through each of the convex hull vertices and has the same shape as the target end face is obtained, which is used as the fitted graph of the target end face;

[0049] In one possible implementation, the calibration component is a disk.

[0050] Fourthly, embodiments of this application provide an electronic device, including:

[0051] Memory, used to store computer programs;

[0052] When the processor executes the program stored in the memory, it implements the steps of the aforementioned articulated robotic arm calibration method.

[0053] Fifthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the aforementioned articulated robotic arm calibration method.

[0054] The articulated robotic arm calibration method provided in this invention involves moving a calibration component to different calibration points using an articulated robotic arm. For each calibration point, point cloud data corresponding to that calibration point is acquired via radar. The approximate position of the calibration point in the radar coordinate system is determined based on the coarse coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system. Then, based on the approximate position of the calibration point and the size of the calibration component, point cloud data belonging to the region where the calibration component is located is selected from the original point cloud data corresponding to that calibration point. This selected point cloud data is then fitted according to the shape of the calibration component to obtain the position of the feature points in the calibration component in the radar coordinate system. Based on the obtained positions and the relative positional relationship between the calibration point and the feature points, the fine position of the calibration point in the radar coordinate system is obtained. Finally, based on the fine positions of each calibration point and their positions in the robotic arm coordinate system, the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system is determined. By using the method of this application embodiment, the fine position of each calibration point in the radar coordinate system is determined by moving the calibration component, thereby accurately determining the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system. This enables the loading and unloading robot to successfully grasp goods and improve the accuracy of grasping and stacking goods.

[0055] Of course, implementing any product or method of the present invention does not necessarily require achieving all of the advantages described above at the same time. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these drawings.

[0057] Figure 1 Example diagram of the loading and unloading robot provided in the embodiments of this application;

[0058] Figure 2 A schematic flowchart illustrating the articulated robotic arm calibration method provided in this application embodiment;

[0059] Figure 3 A schematic diagram showing the installation positions of the radar and robotic arm provided in an embodiment of this application;

[0060] Figure 4a Example diagram of raw point cloud data obtained by the radar scanning calibration component provided in the embodiments of this application;

[0061] Figure 4b A front view of the filtered point cloud data provided in the embodiments of this application;

[0062] Figure 4c A top view of the filtered point cloud data provided in the embodiments of this application;

[0063] Figure 5a A front view of the fitted point cloud data corresponding to the calibration points provided in this application embodiment;

[0064] Figure 5b A side view of the fitted point cloud data corresponding to the calibration points provided in this application embodiment;

[0065] Figure 6 A schematic diagram illustrating the determination of the fine position of the calibration point as provided in an embodiment of this application;

[0066] Figure 7 An example diagram of the projected point cloud data of the plane containing the disk provided in an embodiment of this application;

[0067] Figure 8a This is a front view of the convex hull vertex provided in an embodiment of this application;

[0068] Figure 8b A side view of a convex hull vertex provided in an embodiment of this application;

[0069] Figure 9 This is an example diagram of fitting convex hull vertices provided in an embodiment of this application;

[0070] Figure 10An example diagram showing the position of the fitted graphic in the original point cloud data for an embodiment of this application;

[0071] Figure 11 This is a schematic diagram of the jointed robotic arm calibration device provided in an embodiment of this application. Detailed Implementation

[0072] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art based on this application are within the scope of protection of the present invention.

[0073] With the increasing intelligence of industry, robots are being used more and more widely in production, manufacturing, warehousing, and logistics. Mobile loading and unloading robots are even gradually replacing manual handling. Mobile loading and unloading robots typically consist of a chassis and a robotic arm equipped with a gripper. The chassis is equipped with tracks and wheels. In current technology, loading and unloading robots rely on a LiDAR mounted on the mobile chassis for navigation and positioning, and a camera mounted on the robotic arm for object recognition and operation. Therefore, visual calibration is limited to the relationship between the mobile chassis and the LiDAR, and hand-eye calibration between the robotic arm and the camera. However, loading and unloading robots need to use LiDAR to accurately locate the position of the vehicle, cargo, obstacles, and other work environments and target objects, and then use the robotic arm to precisely grasp the target objects. Therefore, accurate visual calibration between the robotic arm and the LiDAR mounted on the mobile chassis is crucial. Inaccurate calibration can lead to low robotic arm accuracy or even grasping failure.

[0074] In order to accurately determine the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system, and thus accurately grasp and stack goods, this application provides a method, device and automatic calibration system for calibrating an articulated robotic arm. The articulated robotic arm calibration method provided in this application will be described in detail below through specific embodiments.

[0075] In a first aspect, this application provides a method for calibrating an articulated robotic arm, applicable to loading and unloading robots, and also to any electronic device other than loading and unloading robots that possesses articulated robotic arm calibration capabilities. In specific applications, when applied to any electronic device other than loading and unloading robots that possesses articulated robotic arm calibration capabilities, the electronic device can be a server or a terminal device; when applied to loading and unloading robots, the loading and unloading robot can be... Figure 1 The loading and unloading robot 1 shown includes an articulated robotic arm 11, a calibration assembly 12, a mobile chassis 13, and a radar 14.

[0076] The articulated robotic arm 11 is used to grasp and stack goods and has multiple axis joints; the calibration component 12 is installed on the flange at the end of the articulated robotic arm 11; the mobile chassis 13 is used to drive the articulated robotic arm 11 to move so that the articulated robotic arm 11 can grasp and stack goods at different positions; the radar 14 is used to scan and obtain three-dimensional point cloud data such as distance and position of the environment around the loading and unloading robot.

[0077] The following is a detailed description of a jointed robotic arm calibration method provided in the embodiments of this application.

[0078] See Figure 2 , Figure 2 This is a flowchart illustrating a first articulated robotic arm calibration method provided in an embodiment of this application. The method includes:

[0079] Step S1: For multiple different calibration points in the coordinate system of the articulated robotic arm, obtain the point cloud data obtained by the radar scanning calibration component when the end of the articulated robotic arm is at the calibration point, and use it as the original point cloud data corresponding to the calibration point.

[0080] The calibration component is located at the end effector, which is the end of the articulated robotic arm used to grasp goods. The end effector of the articulated robotic arm can be equipped with gripping components such as suction cups or grippers, or it can be equipped with the calibration component. For example, when the articulated robotic arm needs to grasp goods, a gripping component can be installed at the end effector; when calibration is required, the gripping component is removed, and the calibration component is installed at the end effector. Alternatively, both the gripping component and the calibration component can be installed at the end effector of the articulated robotic arm simultaneously; the specific installation method depends on the structure of the end effector.

[0081] The calibration component has an end face for receiving photoelectric signals emitted by the radar. The radar is positioned at either end of the loading and unloading robot, near or far from the picking position. The point cloud data obtained by the radar scanning the calibration component is a dataset of spatial points obtained by the radar scan, and each point cloud contains a three-dimensional coordinate and laser reflection intensity.

[0082] The radar can be any type of lidar capable of scanning and obtaining three-dimensional point cloud data, such as multi-line lidar, solid-state lidar, or semi-solid-state lidar. This application does not limit the scope of the lidar.

[0083] In other embodiments, three-dimensional vision devices such as three-dimensional cameras or three-dimensional video cameras can be used instead of radar to capture three-dimensional point cloud data of the environment around the loading and unloading robot.

[0084] A calibration point is any number of points that are both within the working range of the articulated robotic arm and within the scanning range of the radar. The coordinates of the calibration point are the coordinates in the robotic arm coordinate system.

[0085] To facilitate the calculation of the approximate coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system, in one possible implementation, the end of the loading and unloading robot equipped with the radar can be designated as the front of the robot, the opposite direction as the rear, the directions perpendicular to the robot's front-rear direction as the left and right sides, and the directions perpendicular to the plane on which the robot is located as the top and bottom. The robotic arm coordinate system is a spatial rectangular coordinate system OXYZ, with the origin at the center of the articulated robotic arm's base. The X-direction of the coordinate system is consistent with the robot's front-rear direction, the Y-direction is consistent with the robot's left-right direction, and the Z-direction is consistent with the robot's up-down direction.

[0086] Understandably, in order for the calibration component to receive a sufficient amount of photoelectric signals transmitted by the radar, at each calibration point, the end face of the calibration component used to receive the photoelectric signals transmitted by the radar should face backward.

[0087] Step S2: For each calibration point, determine the approximate position of the calibration point in the radar coordinate system based on the approximate coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system.

[0088] It is understandable that, based on the mechanical mechanism of the loading and unloading robot, a rough coordinate transformation relationship can be obtained between the robot arm coordinate system and the radar coordinate system, thus obtaining the approximate relative position and orientation of the radar mounting location and the center of the articulated robot arm base. For example, such as... Figure 3 As shown, radar A is located at the front of the loading and unloading robot, and the center B of the base of the articulated robotic arm is located on the right side of the loading and unloading robot. According to the mechanical mechanism of the loading and unloading robot, the front-to-back distance between the radar and the articulated robotic arm can be determined as y, and the left-to-right distance as x. The center of the base of the articulated robotic arm is located at an angle α to the right and rear of the radar. Therefore, the coarse coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system can be determined according to the mechanical mechanism of the loading and unloading robot. By converting the coordinates of the calibration point in the robotic arm coordinate system to the coarse coordinates in the radar coordinate system, the approximate position of the calibration point in the radar coordinate system can be obtained.

[0089] In order to perform accurate calibration, in step S3, for each calibration point, based on the approximate location of the calibration point and the size of the calibration component, the point cloud data belonging to the region where the calibration component is located is filtered out from the original point cloud data corresponding to the calibration point, and used as the filtered point cloud data corresponding to the calibration point.

[0090] It is understandable that filtering point cloud data belonging to the area where the calibration component is located from the original point cloud data corresponding to the calibration point can be done by filtering point cloud data within a preset distance from the calibration component from the original point cloud data corresponding to the calibration point. For example, the preset distance can be set to 3cm, so point cloud data within 3cm of the calibration component can be used as the filtered point cloud data corresponding to that calibration point.

[0091] It is understandable that the shape of the filtered point cloud data is related to the shape of the calibration component. In other words, the shape of the filtered point cloud data is related to the shape of the end face of the calibration component used to receive the photoelectric signals emitted by the radar. For example, if the end face of the calibration component used to receive the photoelectric signals emitted by the radar is circular, Figure 4a The image shows the original point cloud data obtained by the radar scanning calibration component when the end effector of the articulated robotic arm is at the calibration point in step S1. The front and top views of the filtered point cloud data are shown below. Figure 4b and Figure 4c As shown, the filtered point cloud data is the point cloud data around the calibration component, and the circular part in the front view of the filtered point cloud data is the point cloud data emitted onto the calibration component.

[0092] In one possible implementation, when filtering point cloud data belonging to the region where the calibration component is located in step S3 above, filtering can also be performed by setting a bounding box with a size larger than the calibration component, that is, using the point cloud data within the bounding box as the filtered point cloud data. Therefore, step S3 above may include:

[0093] Step S31: Based on the size of the calibration component, set a bounding box at a rough location that can accommodate the calibration component;

[0094] Step S32: Select the point cloud data located within the bounding box from the original point cloud data corresponding to the calibration point, and use it as the filtered point cloud data corresponding to the calibration point.

[0095] The size of the enclosure should be larger than the size of the calibration component. For example, if the calibration component is a disc with a diameter of 200 mm and a thickness of 10 mm, the enclosure can be a cube with a side length of 400 mm.

[0096] Understandably, the bounding box should be large enough to accommodate the calibration components, but not excessively large. Specifically, a preset threshold can be set for the size of the bounding box; for example, the distance between the edge of the bounding box closest to the calibration components and the calibration components can be set to 3–5 cm.

[0097] The shape of the bounding box can be a cube, a cylinder, or a sphere; the specific shape can be set according to the actual situation.

[0098] By setting a bounding box that can accommodate the calibration component, the point cloud data around the calibration component can be automatically filtered. This method is more convenient, faster, and more accurate than manually filtering point cloud data, and it can ensure that all point cloud data around the calibration component is filtered out.

[0099] To facilitate subsequent processing and analysis, after obtaining the filtered point cloud data, in step S4, for each calibration point, the filtered point cloud data corresponding to the calibration point is fitted according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system. Based on the obtained position and the relative positional relationship between the calibration point and the feature point, the fine position of the calibration point in the radar coordinate system is obtained.

[0100] In this context, the feature points in the calibration component are easily identifiable points in the fitted graph obtained by fitting the shape of the end face of the calibration component used to receive the photoelectric signal transmitted by the radar. For example, if the calibration component is a cylinder and the end face of the calibration component used to receive the photoelectric signal transmitted by the radar is circular, the fitted graph is a circle, and the feature point is the center of the circle.

[0101] Fitting the filtered point cloud data corresponding to the calibration points according to the shape of the calibration component essentially involves fitting the complex filtered point cloud data into a planar model corresponding to the shape of the calibration component, facilitating subsequent processing and analysis. The RANSAC (Random Sample Consensus) algorithm is typically used, which finds the optimal planar model by randomly sampling a subset from the point cloud data and estimating the error between the planar model and the sampling points. Principal component analysis can also be used to fit the point cloud data into a plane. This application does not limit the fitting method.

[0102] It is understandable that fitting the filtered point cloud data corresponding to the calibration point according to the shape of the calibration component actually means fitting the filtered point cloud data corresponding to the calibration point according to the shape of the end face of the calibration component used to receive the photoelectric signal transmitted by the radar. For example, if the end face of the calibration component used to receive the photoelectric signal transmitted by the radar is circular, then the front view and side view of the filtered point cloud data corresponding to the calibration point are fitted according to the circle as shown below. Figure 5a and Figure 5b As shown, the shape of the filtered point cloud data corresponding to the fitted calibration point is consistent with the shape of the calibration component, which is circular, and the fitted data is in a plane. Then the position of the center of the circle in the radar coordinate system can be determined. Since the positional relationship between the center of the circle and the calibration point can be obtained in advance, the fine position of the calibration point in the radar coordinate system can be determined based on the position of the center of the circle and the positional relationship between the center of the circle and the calibration point.

[0103] After repeating steps S1-S4 to obtain the fine position of each calibration point, step S5 determines the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system based on the fine position of each calibration point and its position in the robotic arm coordinate system.

[0104] Specifically, equations relating the robotic arm coordinate system and the radar coordinate system can be established using a coordinate transformation matrix. Solving these equations yields a precise coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system. For example, equation (M) can be established using a coordinate transformation matrix M. (4×4) ·(P radar ) (4×n) =(P robot ) (4×n) , where P radar It is a 4xn matrix, where the i-th column represents the homogeneous coordinates of the i-th calibration point in the radar coordinate system, denoted by (x... i ,y i ,z i ,1) indicates that the k-th row represents the value of the k-th component of the homogeneous coordinates, where n is the number of calibration points, and P robot It is a 4xn matrix, where the i-th column represents the homogeneous coordinate value of the i-th calibration point in the robot's coordinate system, denoted by (x... i ,y i ,z i ,1) indicates that the k-th row represents the value of the k-th component of the homogeneous coordinates, which is obtained by solving the equation M=P robot ·P radar T ·(P radar ·P radar T ) -1 This allows us to obtain the fine coordinate transformation relationship M between the robotic arm coordinate system and the radar coordinate system.

[0105] By using the method of this application embodiment, the fine position of each calibration point in the radar coordinate system is determined by moving the calibration component, thereby accurately determining the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system. This enables the loading and unloading robot to successfully grasp goods and improve the accuracy of grasping and stacking goods.

[0106] It is understandable that, in step S31 above, when setting a bounding box capable of accommodating the calibration component based on its dimensions, if the shape of the bounding box is set to a cube, then step S31 can specifically be as follows:

[0107] Based on the maximum spatial span of the calibrated components, a cube with a side length greater than the maximum spatial span is set as the bounding box at a rough location.

[0108] For example, if the maximum spatial span of the calibrated component is 2 cm, then the side length of the bounding box should be greater than 2 cm.

[0109] The method of this application embodiment sets the size of the bounding box according to the maximum spatial span of the calibration component, so that the point cloud data screened by the bounding box includes all point cloud data around the calibration component, avoiding omissions.

[0110] It is understandable that the calibration component can be fixed to the end of the articulated robotic arm via a connecting component, which may include flanges, screws, bolts, etc., or it may be fixed to the end of the articulated robotic arm by welding or other means.

[0111] To ensure that the radar cannot scan the flange and thus avoid interfering with the point cloud data of the calibration component, in one possible implementation, the size of the target end face is larger than the size of the flange. It is understood that the larger the size of the target end face, the more data it receives. Even when the calibration component is far from the radar, the point cloud data on the target end face of the calibration component can still meet the fitting conditions, expanding the range of calibration point settings and thus improving the robustness of the calibration.

[0112] In one possible implementation, if the calibration component is connected to the end of the articulated robotic arm via a flange, the calibration component has a target end face and a fixed end face. The target end face is the end face of the calibration component that is furthest from the flange, and the fixed end face of the calibration component is fixed to the flange at the end. The normal of the target end face of the calibration component is collinear with the normal of the flange, the feature point is the center of the target end face of the calibration component, and the size of the target end face is larger than the size of the flange. The target end face is used to receive photoelectric signals transmitted by the radar.

[0113] Therefore, step S4 above may specifically include:

[0114] Step S41: Fit the filtered point cloud data corresponding to the calibration points according to the shape of the target end face to obtain the fitted graphic of the target end face;

[0115] In one possible implementation, in step S4 above, the filtered point cloud data is fitted to obtain multiple fitting planes. From these multiple fitting planes, the plane with the smallest root mean square error and closest to the origin of the radar coordinate system is selected to obtain the point cloud data of the target end face, thereby obtaining the fitted graphic of the target end face.

[0116] In one possible implementation, after step S41, in order to make the filtered point cloud data more refined, clustering and noise reduction processes can be performed on the filtered point cloud data in step S41.

[0117] Step S42: Based on the fitted graph, determine the position and normal of the center of the target end face in the radar coordinate system;

[0118] Since the fixed end face of the calibration component is fixed to the flange at the end, and the normal of the target end face of the calibration component is collinear with the normal of the flange, in step S43, the fine position of the calibration point in the radar coordinate system can be calculated based on the thickness of the calibration component, the obtained position and normal, where the thickness is the spatial span of the calibration component along the central axis.

[0119] For example, if the position of the center of the target end face in the radar coordinate system is determined according to step S42, such as... Figure 6 Point A in the middle, and the normal direction is as follows Figure 6 As shown, if the thickness of the calibration component is 10mm, then the precise position of the calibration point is achieved by shifting point A 10mm along the normal direction. Figure 6 Midpoint B.

[0120] Using the method of this application embodiment, the filtered point cloud data corresponding to the calibration point is fitted according to the shape of the target end face of the calibration component to obtain the fitted graphic of the target end face. Then, the position and normal of the center of the target end face in the radar coordinate system can be determined. Since the normal of the target end face of the calibration component is collinear with the normal of the flange, and the thickness of the calibration component is known, the fine position of the calibration point in the radar coordinate system can be calculated by conventional calculation methods based on the thickness of the calibration component, the obtained position and normal. The calculation is simple and fast.

[0121] To make the fitted image more accurate, in one possible implementation, the point cloud data can be projected and the convex hull of the point cloud data can be extracted. Therefore, step S41 above includes:

[0122] Step S411: Determine the convex hull vertices in the filtered point cloud;

[0123] Step S412: Based on each convex hull vertex, fit a fitted graph that passes through each convex hull vertex and has the same shape as the target end face, and use it as the fitted graph of the target end face.

[0124] Determining the convex hull vertices in the filtered point cloud essentially involves projecting the filtered point cloud data onto a plane and then identifying the vertices that form the convex hull from the polygon formed by the outermost points. These vertices are then connected in a specific order to form the edge of the convex hull. Specifically, the Graham scan method or the Open3D tool library can be used to determine the convex hull vertices in the filtered point cloud.

[0125] It is understandable that, since the size of the target end face is larger than the size of the flange, the shape of the fitted image obtained by fitting each convex hull vertex is the same as the shape of the target end face. Therefore, the fitted image obtained by fitting each convex hull vertex and whose shape is the same as the shape of the target end face is the fitted image of the target end face.

[0126] Using the method of this application embodiment, convex hull vertices are determined in the filtered point cloud, and a fitting graph that passes through each convex hull vertex and has the same shape as the target end face is obtained based on each convex hull vertex. This fitting graph is used as the fitting graph of the target end face, making the obtained fitting graph of the target end face more accurate.

[0127] To improve computational efficiency, in one possible implementation, the calibration component is a disk.

[0128] In one possible embodiment, after fitting the filtered point cloud data corresponding to the calibration points according to step S41 above to obtain a circular fitted shape for the target end face, the point cloud data of the disk surface can be fitted again, and then the point cloud data of the disk surface can be projected onto the plane, such as... Figure 7 The projection point cloud data of the plane containing the disk is obtained as shown. Then, the convex hull in the projection point cloud data is calculated and the convex hull vertices are obtained. The front view and side view of the convex hull vertices are shown below. Figure 8a and Figure 8b As shown, it is roughly circular, and then... Figure 9 As shown, a circle is fitted based on each convex hull vertex to obtain the fitted graph corresponding to the disk. Finally, the position and normal of the circle's center in the radar coordinate system are determined through the above step S42. The final position of the fitted circle in the original data point cloud is shown in the figure. Figure 10 As shown.

[0129] By using a disc-shaped calibration component, compared to calibration components of other shapes, the point cloud data on the surface of the disc can be quickly filtered out from the original point cloud data, and the coordinates of the center of the disc in the radar coordinate system can be quickly calculated. This improves the efficiency of calculating the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system.

[0130] Understandably, in one possible implementation, when the loading and unloading robot is equipped with two articulated robotic arms, a calibration component is installed at the end of each of the two articulated robotic arms. The articulated robotic arms are then calibrated according to the method described above to obtain the fine coordinate transformation relationship between the coordinate systems of the two robotic arms and the radar coordinate system. When the loading and unloading robot with two articulated robotic arms is used to grasp and stack goods, the calibration results of the two articulated robotic arms are used for positioning.

[0131] Secondly, embodiments of this application provide an automatic calibration system, which includes an articulated robotic arm, a radar, and a processor, with a calibration component provided at the end of the articulated robotic arm;

[0132] The processor controls the movement of the articulated robotic arm so that the end effector moves sequentially to multiple different calibration points in the robotic arm coordinate system; and whenever the end effector moves to a new calibration point, it sends a scan request to the radar, wherein each calibration point is within the radar's field of view;

[0133] The radar is used to scan in response to a scan request, obtain point cloud data, and send the point cloud data to the processor;

[0134] The processor is also used to: acquire point cloud data obtained by scanning the calibration component with radar when the end effector of the articulated robotic arm is at the calibration point, as the original point cloud data corresponding to the calibration point; determine the approximate position of the calibration point in the radar coordinate system based on the coarse coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system for each calibration point; filter point cloud data belonging to the region where the calibration component is located from the original point cloud data corresponding to the calibration point based on the approximate position of the calibration point and the size of the calibration component, as the filtered point cloud data corresponding to the calibration point; fit the filtered point cloud data corresponding to the calibration point according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system, and obtain the fine position of the calibration point in the radar coordinate system based on the obtained position and the relative positional relationship between the calibration point and the feature point; and determine the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system based on the fine position of each calibration point and the position of each calibration point in the robotic arm coordinate system.

[0135] Through the embodiments of this application, by moving the calibration component, the precise position of each calibration point in the radar coordinate system can be determined at each calibration point. This allows for the accurate determination of the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system, thereby enabling the loading and unloading robot to successfully grasp goods and improve the accuracy of the loading and unloading robot in grasping and stacking goods.

[0136] To more clearly illustrate the articulated robotic arm calibration method of this application, the following description is provided in conjunction with specific embodiments.

[0137] Taking a disk as the calibration component as an example, multiple calibration points are randomly selected within the working range of the articulated robotic arm and the scanning range of the radar. The articulated robotic arm moves the disk to the first calibration point, with the disk surface facing backward. After scanning, the radar obtains the original point cloud data corresponding to that calibration point, such as... Figure 4aAs shown, the approximate position and orientation of the radar and the center of the articulated robotic arm base are obtained through the known mechanical mechanisms of the loading and unloading robot. The coordinates of the current calibration point in the robotic arm coordinate system are converted into a coarse coordinate in the radar coordinate system, thus obtaining the coarse coordinates of the disk in the radar coordinate system. Then, the point cloud data around the disk is filtered using a cube with a side length greater than the maximum size of the disk space, retaining all the point cloud data within the cube. The filtered point cloud data is shown below. Figure 4b As shown, the filtered point cloud data is then fitted. From the multiple fitted planes obtained, the plane with the smallest root mean square error and the closest to the origin of the radar coordinate system is selected, thus obtaining the point cloud data of the disk surface, as shown. Figure 5a As shown.

[0138] Then, the point cloud data on the disk surface is fitted again to obtain the plane on which the disk lies, and the point cloud data on the disk surface is projected onto this plane to obtain the projected point cloud data, as shown below. Figure 7 As shown, the convex hull in the projected point cloud data is then calculated and its vertices are obtained, as follows: Figure 8a As shown, and as Figure 9 As shown, by fitting a circle based on the vertices of the convex hull, the normal, radius, and center coordinates of the circular plane can be obtained.

[0139] Repeat the above steps for each calibration point to obtain the precise coordinates of that calibration point in the radar coordinate system. Finally, establish equation (M) using the transformation matrix M to obtain the coordinates of the center of the disk corresponding to each calibration point. (4×4) ·(P radar ) (4×n) =(P robot ) (4×n) By finding the least squares solution M = P to this equation robot ·P radar T ·(P radar ·p radar T ) -1 This allows us to obtain the transformation relationship M between the articulated robotic arm coordinate system and the lidar coordinate system, thus completing the calibration process.

[0140] Thirdly, embodiments of this application provide a calibration device for an articulated robotic arm, such as... Figure 11 As shown, the device includes:

[0141] The data acquisition module 1101 is used to acquire point cloud data obtained by radar scanning calibration component when the end of the articulated robot arm is at multiple different calibration points in the robot arm coordinate system, and use it as the original point cloud data corresponding to the calibration point. The calibration component is set at the end.

[0142] The coarse position determination module 1102 is used to determine the coarse position of each calibration point in the radar coordinate system based on the coarse coordinate transformation relationship between the robot arm coordinate system and the radar coordinate system.

[0143] The data filtering module 1103 is used to filter out the point cloud data belonging to the area where the calibration component is located from the original point cloud data corresponding to each calibration point, based on the approximate location of the calibration point and the size of the calibration component, and use it as the filtered point cloud data corresponding to the calibration point.

[0144] The fine position determination module 1104 is used to fit the filtered point cloud data corresponding to each calibration point according to the shape of the calibration component, obtain the position of the feature point in the calibration component in the radar coordinate system, and obtain the fine position of the calibration point in the radar coordinate system based on the obtained position and the relative positional relationship between the calibration point and the feature point.

[0145] The relationship determination module 1105 is used to determine the fine coordinate transformation relationship between the robot arm coordinate system and the radar coordinate system based on the fine position of each calibration point and the position of each calibration point in the robot arm coordinate system.

[0146] Through the embodiments of this application, by moving the calibration component, the precise position of each calibration point in the radar coordinate system can be determined at each calibration point. This allows for the accurate determination of the coordinate transformation relationship between the radar coordinate system and the robotic arm coordinate system, thereby enabling the loading and unloading robot to successfully grasp goods and improve the accuracy of the loading and unloading robot in grasping and stacking goods.

[0147] In one possible implementation, the data filtering module 1103 includes:

[0148] The bounding box setting submodule is used to set a bounding box that can accommodate the calibration component at a rough location, based on the size of the calibration component;

[0149] The data filtering submodule is used to filter out the point cloud data located within the bounding box from the original point cloud data corresponding to the calibration point, and use it as the filtered point cloud data corresponding to the calibration point.

[0150] In one possible implementation, the bounding box setting submodule is specifically used for:

[0151] Based on the maximum spatial span of the calibrated components, a cube with a side length greater than the maximum spatial span is set as the bounding box at a rough location;

[0152] The calibration assembly has a target end face and a fixed end face. The target end face is the end face of the calibration assembly that is farthest from the flange. The fixed end face of the calibration assembly is fixed to the flange at the end. The normal of the target end face of the calibration assembly is collinear with the normal of the flange. The feature point is the center of the target end face of the calibration assembly. The size of the target end face is larger than the size of the flange.

[0153] In one possible implementation, the fine position determination module 1104 includes:

[0154] The data fitting submodule is used to fit the filtered point cloud data corresponding to the calibration points according to the shape of the target end face to obtain the fitted graphic of the target end face.

[0155] The normal determination submodule is used to determine the position and normal of the center of the target end face in the radar coordinate system based on the fitted graph;

[0156] The position calculation submodule is used to calculate the fine position of the calibration point in the radar coordinate system based on the thickness of the calibration component, the obtained position and normal, where the thickness is the spatial span of the calibration component along the central axis.

[0157] In one possible implementation, the data fitting submodule is specifically used for:

[0158] Determine the convex hull vertices in the filtered point cloud;

[0159] Based on each convex hull vertex, a fitted graph that passes through each convex hull vertex and has the same shape as the target end face is obtained, which is used as the fitted graph of the target end face.

[0160] In one possible implementation, the calibration component is a disk.

[0161] Fourthly, embodiments of this application provide an electronic device, including:

[0162] Memory, used to store computer programs;

[0163] When a processor executes a program stored in memory, it performs the following steps:

[0164] For multiple different calibration points in the coordinate system of the articulated robotic arm, point cloud data obtained by radar scanning calibration components when the end effector of the articulated robotic arm is located at the calibration points are obtained, and these are used as the original point cloud data corresponding to the calibration points. The calibration components are set at the end effector.

[0165] For each calibration point, the approximate position of the calibration point in the radar coordinate system is determined based on the coarse coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system.

[0166] For each calibration point, based on the approximate location of the calibration point and the size of the calibration component, the point cloud data belonging to the region where the calibration component is located is filtered out from the original point cloud data corresponding to the calibration point, and used as the filtered point cloud data corresponding to the calibration point.

[0167] For each calibration point, the filtered point cloud data corresponding to the calibration point is fitted according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system. Based on the obtained position and the relative positional relationship between the calibration point and the feature point, the fine position of the calibration point in the radar coordinate system is obtained.

[0168] Based on the precise positions of each calibration point and their positions in the robotic arm coordinate system, a precise coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system is determined.

[0169] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0170] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0171] In another embodiment of the present invention, a computer-readable storage medium is also provided, which stores a computer program that, when executed by a processor, implements the steps of any of the above-described joint robotic arm calibration methods.

[0172] In another embodiment of the present invention, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the joint robotic arm calibration methods described in the above embodiments.

[0173] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0174] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0175] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system and device embodiments are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0176] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A method for calibrating an articulated robotic arm, characterized in that, The method includes: For multiple different calibration points in the coordinate system of the articulated robotic arm, point cloud data obtained by radar scanning calibration components when the end effector of the articulated robotic arm is located at the calibration points are obtained, and these are used as the original point cloud data corresponding to the calibration points. The calibration components are set at the end effector. For each calibration point, the approximate position of the calibration point in the radar coordinate system is determined based on the coarse coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system. For each calibration point, based on the approximate location of the calibration point and the size of the calibration component, the point cloud data belonging to the region where the calibration component is located is filtered out from the original point cloud data corresponding to the calibration point, and used as the filtered point cloud data corresponding to the calibration point. For each calibration point, the filtered point cloud data corresponding to the calibration point is fitted according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system. Based on the obtained position and the relative positional relationship between the calibration point and the feature point, the fine position of the calibration point in the radar coordinate system is obtained. Based on the precise positions of each calibration point and their positions in the robotic arm coordinate system, a precise coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system is determined.

2. The method according to claim 1, characterized in that, The step of filtering point cloud data belonging to the region where the calibration component is located from the original point cloud data corresponding to the calibration point, based on the approximate location of the calibration point and the size of the calibration component, as the filtered point cloud data corresponding to the calibration point, includes: Based on the dimensions of the calibration component, a surrounding box capable of accommodating the calibration component is provided at the approximate location; Point cloud data located within the bounding box are selected from the original point cloud data corresponding to the calibration point and used as the filtered point cloud data corresponding to the calibration point.

3. The method according to claim 2, characterized in that, The step of setting a surrounding box capable of accommodating the calibration component at the approximate location, based on the size of the calibration component, includes: Based on the maximum spatial span of the calibrated components, a cube with a side length greater than the maximum spatial span is set as a bounding box at the approximate location.

4. The method according to claim 1, characterized in that, The calibration component has a target end face and a fixed end face. The fixed end face of the calibration component is fixed to the flange at the end. The target end face is the end face of the calibration component that is away from the flange. The normal of the target end face of the calibration component is collinear with the normal of the flange. The feature point is the center of the target end face of the calibration component. The size of the target end face is larger than the size of the flange. The step of fitting the filtered point cloud data corresponding to the calibration point according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system, and obtaining the fine position of the calibration point in the radar coordinate system based on the obtained position and the relative positional relationship between the calibration point and the feature point, includes: The filtered point cloud data corresponding to the calibration points are fitted according to the shape of the target end face to obtain the fitted graphic of the target end face; Based on the fitted graph, determine the position and normal of the center of the target end face in the radar coordinate system; Based on the thickness of the calibration component, the obtained position, and the normal, the fine position of the calibration point in the radar coordinate system is calculated, wherein the thickness is the spatial span of the calibration component along the central axis.

5. The method according to claim 4, characterized in that, The step of fitting the filtered point cloud data corresponding to the calibration points according to the shape of the target end face to obtain the fitted graphic of the target end face includes: Determine the convex hull vertices in the filtered point cloud; Based on each of the convex hull vertices, a fitted graph that passes through each of the convex hull vertices and has the same shape as the target end face is obtained, and this fitted graph is used as the fitted graph of the target end face.

6. The method according to claim 4, characterized in that, The calibration component is a disk.

7. An automatic calibration system, characterized in that, The system includes an articulated robotic arm, a radar, and a processor, with a calibration component at the end of the articulated robotic arm. The processor is configured to control the movement of the articulated robotic arm so that the end effector sequentially moves to multiple different calibration points in the robotic arm coordinate system of the articulated robotic arm; and whenever the end effector moves to a new calibration point, it sends a scan request to the radar, wherein each calibration point is located within the field of view of the radar; The radar is configured to perform a scan in response to the scan request, obtain point cloud data, and send the point cloud data to the processor; The processor is further configured to: acquire point cloud data obtained by scanning the calibration component with the radar when the end effector of the articulated robotic arm is at the calibration point, for each calibration point, as the original point cloud data corresponding to the calibration point; determine the approximate position of the calibration point in the radar coordinate system based on the approximate coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system for each calibration point; and filter out data belonging to the calibration component from the original point cloud data corresponding to the calibration point based on the approximate position of the calibration point and the size of the calibration component for each calibration point. The point cloud data of the specified area is used as the filtered point cloud data corresponding to the calibration point. For each calibration point, the filtered point cloud data corresponding to the calibration point is fitted according to the shape of the calibration component to obtain the position of the feature point in the calibration component in the radar coordinate system. Based on the obtained position and the relative positional relationship between the calibration point and the feature point, the fine position of the calibration point in the radar coordinate system is obtained. Based on the fine position of each calibration point and the position of each calibration point in the robotic arm coordinate system, the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system is determined.

8. A calibration device for an articulated robotic arm, characterized in that, The device includes: The data acquisition module is used to acquire point cloud data obtained by a radar scanning calibration component when the end of the articulated robotic arm is located at multiple different calibration points in the coordinate system of the articulated robotic arm, and use the calibration component as the original point cloud data corresponding to the calibration point. The calibration component is set at the end. The coarse position determination module is used to determine the coarse position of each calibration point in the radar coordinate system based on the coarse coordinate transformation relationship between the robot arm coordinate system and the radar coordinate system. The data filtering module is used to filter out point cloud data belonging to the region where the calibration component is located from the original point cloud data corresponding to each calibration point, based on the approximate location of the calibration point and the size of the calibration component, and use it as the filtered point cloud data corresponding to the calibration point. The fine position determination module is used to fit the filtered point cloud data corresponding to each calibration point according to the shape of the calibration component, to obtain the position of the feature point in the calibration component in the radar coordinate system, and to obtain the fine position of the calibration point in the radar coordinate system based on the obtained position and the relative positional relationship between the calibration point and the feature point. The relationship determination module is used to determine the fine coordinate transformation relationship between the robotic arm coordinate system and the radar coordinate system of the radar based on the fine position of each calibration point and the position of each calibration point in the robotic arm coordinate system.

9. The apparatus according to claim 8, characterized in that, The data filtering module includes: A bounding box setting submodule is used to set a bounding box capable of accommodating the calibration component at the coarse location, based on the size of the calibration component; The data filtering submodule is used to filter out the point cloud data located within the bounding box from the original point cloud data corresponding to the calibration point, and use it as the filtered point cloud data corresponding to the calibration point. The bounding box setting submodule is specifically used for: Based on the maximum spatial span of the calibration component, a cube with a side length greater than the maximum spatial span is set as a bounding box at the approximate location; The calibration component has a target end face and a fixed end face. The fixed end face of the calibration component is fixed to the flange at the end. The target end face is the end face of the calibration component that is away from the flange. The normal of the target end face of the calibration component is collinear with the normal of the flange. The feature point is the center of the target end face of the calibration component. The size of the target end face is larger than the size of the flange. The fine position determination module includes: The data fitting submodule is used to fit the filtered point cloud data corresponding to the calibration point according to the shape of the target end face to obtain the fitted graphic of the target end face. The normal direction determination submodule is used to determine the position and normal of the center of the target end face in the radar coordinate system based on the fitted graph. The position calculation submodule is used to calculate the fine position of the calibration point in the radar coordinate system based on the thickness of the calibration component, the obtained position and normal, wherein the thickness is the spatial span of the calibration component along the central axis direction; The data fitting submodule is specifically used for: Determine the convex hull vertices in the filtered point cloud; Based on each of the convex hull vertices, a fitted graph that passes through each of the convex hull vertices and has the same shape as the target end face is obtained, which is used as the fitted graph of the target end face; The calibration component is a disk.

10. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the steps of the method described in any one of claims 1-6.

Images