End-to-end mechanical arm guide calibration method and system based on pose detection compensation
By adopting an end-to-end robotic arm guidance and calibration method based on pose detection and compensation in automobile production, and using a 3D profilometer and encoder for vehicle body point cloud data registration and prediction, the problem of accumulated robotic arm calibration error was solved, achieving high-precision robotic arm operation and improved production quality.
Patent Information
- Application Number
- CN202511482043.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-16
- Publication Date
- 2026-02-03
AI Technical Summary
Existing technologies for calibrating robotic arms in automobile production suffer from insufficient accuracy, poor adaptability, and the cumulative effect of multiple calibration errors, leading to inaccurate robotic arm operations.
An end-to-end robotic arm guidance and calibration method based on pose detection and compensation is adopted. By setting multiple vehicle body scanning positions on the production line chain, the point cloud data of the vehicle body is obtained using a 3D profilometer, and registration and spatial transformation are performed. Combined with encoder prediction of vehicle body position changes, an accurate transformation relationship between the robotic arm base coordinate system and the production line chain coordinate system is established, and neural network is used for error compensation.
It improves the operational accuracy and efficiency of robotic arms, reduces calibration costs, reduces human intervention, and improves production quality and safety.
Smart Images
Figure CN121447618A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of robot calibration and calibration technology, and in particular to an end-to-end robotic arm guidance and calibration method and system based on pose detection and compensation. Background Technology
[0002] In automobile production, the production line is highly flexible, with multiple models running on the assembly line. During the process of a vehicle going on the line, the position of the vehicle body relative to the conveyor belt coordinate system changes. When a robotic arm is needed for precise operation, it is necessary to accurately sense the changes in its position and guide and correct the position of the robotic arm, such as for vehicle gap surface difference detection, body welding, painting, etc.
[0003] The current mainstream technical solution involves using four 3D cameras to scan the vehicle body, acquiring point cloud data and obtaining four reference points. Next, the spatial transformation of the vehicle body's center point coordinate system relative to the production line chain coordinate system is calculated, and then converted to the robotic arm's base coordinate system. The calibration process requires calibrating the corresponding 3D cameras to the production line chain coordinate system, and simultaneously calibrating the relationship between the robotic arm's base coordinate system and the chain coordinate system. This involves two calibration steps, requiring the design of high-precision calibration objects. The operation is complex, and the cumulative effect of errors from multiple calibration conversions is significant.
[0004] When a robotic arm operates on a vehicle body, it needs to know the vehicle's position and orientation precisely to ensure that the robotic arm can accurately perform its tasks. However, traditional methods may suffer from insufficient precision and poor adaptability. Summary of the Invention
[0005] The purpose of this invention is to provide an end-to-end robotic arm guidance and calibration method based on pose detection and compensation, which aims to solve the technical problems of the drawbacks of traditional calibration methods.
[0006] To address the above problems, this invention provides an end-to-end robotic arm guidance and calibration method based on pose detection and compensation, the corresponding technical solution of which includes: Step 1: Set multiple vehicle body scanning positions on the production line chain and acquire the corresponding vehicle body point cloud data through the acquisition units set at each vehicle body scanning position; Step 2: Register the vehicle body point cloud data with the standard vehicle body point cloud to obtain the position information of multiple vehicle body reference points; Step 3: Based on the vehicle body reference point position information, obtain the spatial transformation data of the vehicle body center point coordinates relative to the production line plate chain base coordinate system; and obtain the position of the vehicle body center point in the robot arm base coordinate system to obtain the position of the vehicle body center point in the robot arm base coordinate system. Step 4: Install an encoder on the chain drive shaft to obtain the chain speed and predict the position of the vehicle body center point in the robot arm's base coordinate system at the next moment; thus, predict the position change of the vehicle body in real time.
[0007] Furthermore, in step 1, the vehicle body scanning positions are a number of scanning positions evenly distributed around the vehicle body; the acquisition unit uses a 3D profilometer to synchronously trigger a scan of the vehicle body to obtain vehicle body point cloud data.
[0008] Furthermore, step 2 includes: Assume the point cloud data measured by four 3D profilometers are represented as follows: This ensures that each 3D profiler corresponds to a point cloud of a specific area of the vehicle body; the standard vehicle body point cloud is represented as: ; The calibration matrix for the point cloud registration transformation process from the 3D profiler to the corresponding area of the vehicle body is represented as follows: ;in,
[0009] Where p represents a point in the point cloud data measured by the 3D profilometer, p ’ Represented as points on the initial standard vehicle body point cloud; The registration operation is performed by solving the calibration matrix to obtain the position of the vehicle body reference point in the corresponding area. In the vehicle body scanning system, a transformation matrix is used to register the point cloud acquired by the 3D profilometer with the standard vehicle body point cloud in real time and extract reference points; among which, p i ref = T i sensor→std p i sensor .
[0010] In a further embodiment, step 3 described above includes: Step 31: Based on the vehicle body reference point position information, obtain the spatial transformation data of the vehicle body center point coordinates relative to the production line chain coordinate system; specifically, this refers to obtaining the spatial transformation data of the vehicle body center point coordinates relative to the production line chain coordinate system based on the vehicle body reference point position information. Point cloud positions of each standard vehicle body The corresponding spatial transformation data is obtained by solving the rigid body transformation model, and the corresponding solution formula is as follows:
[0011] in: ; Step 32: Obtain the spatial position of the vehicle's center point relative to the robot arm's base coordinate system at the same moment; the corresponding calculation formula is as follows;
[0012] in, This represents the transformation relationship from the robot arm's base coordinate system to the production line's chain coordinate system. The coordinates of the vehicle's center point in the standard vehicle coordinate system, where center represents the center point in the standard vehicle coordinate system; Step 33: Obtain the position of the vehicle body center point in the robot arm's base coordinate system. The corresponding formula is: .
[0013] Furthermore, step 4 above utilizes the chain speed obtained from the encoder. The formula for predicting the position of the vehicle's center point in the robot arm's base coordinate system at the next moment is: .
[0014] Based on the same inventive concept, the present invention also provides an end-to-end robotic arm guidance and calibration system based on pose detection and compensation, comprising: The data acquisition unit is used to set multiple body scanning positions on the production line chain and acquire the corresponding body point cloud data through the acquisition units set at each body scanning position. The data registration unit is used to register the vehicle body point cloud data with the standard vehicle body point cloud to obtain the position information of multiple vehicle body reference points; The data calculation unit is used to obtain the spatial transformation data of the vehicle center point coordinates relative to the production line plate chain base coordinate system based on the vehicle body reference point position information; and to obtain the position of the vehicle center point in the robot arm base coordinate system, so as to obtain the position of the vehicle center point in the robot arm base coordinate system. The data prediction unit is used to install on the chain drive shaft and use an encoder to obtain the chain speed, and predict the position of the vehicle body center point in the robot arm base coordinate system at the next moment; thus, it can predict the position change of the vehicle body in real time.
[0015] Implementing the embodiments of the present invention will have the following beneficial effects: The method and system of this invention can improve the operational accuracy and efficiency of robotic arms, reduce calibration costs, reduce manual intervention, and improve production quality and safety. Specifically, this invention first corrects the robotic arm's trajectory by acquiring vehicle pose information from the moving production line, accurately locating the operation position. Simultaneously, it proposes an effective calibration method: a neural network error compensation strategy based on predicted and actual coordinates. Compared to conventional calibration methods (calibrating the pose camera and production line coordinates, calibrating the robotic arm's base coordinates and production line coordinates, requiring two calibrations), this effectively avoids the accumulation of errors from multiple calibrations. Attached Figure Description
[0016] 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 drawings can be obtained based on these drawings without creative effort.
[0017] in: Figure 1 This is a flowchart illustrating the overall steps of the present invention. Detailed Implementation
[0018] 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 without creative effort are within the scope of protection of the present invention.
[0019] It should be noted that all directional indicators (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.
[0020] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" and "second" may explicitly or implicitly include at least one of the stated features. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.
[0021] This application applies to vehicle body pose measurement and robotic arm guidance scenarios, and current existing technologies have certain problems. For example, because the robotic arm cannot be directly associated with the pose detection camera and there is no common operational field of view (i.e., the robotic arm cannot be directly moved to the same target reference point), direct calibration is not possible. Furthermore, the conventional method is a two-stage secondary calibration, which establishes a connection by separately calibrating the production line coordinates, making the solution complex and difficult to operate. For instance, if existing hand-eye calibration methods are to be applied to vehicle body pose measurement and robotic arm guidance scenarios, the core steps are: first, to obtain a point cloud dataset based on the camera coordinate system; second, to select the corresponding target point cloud calibration point, whose first coordinate relative to the camera coordinate system is already obtained from the first step; third, to drive the robotic arm to move directly to the calibration point (here, the calibration point actually needs to be reachable within the robotic arm's range, and achieving high-precision alignment is not easy to achieve, making this method unsuitable for this scenario); and finally, to directly obtain the second coordinate of the calibration point under the robotic arm's base coordinates.
[0022] To address the drawbacks of the aforementioned technologies, such as Figure 1 The present invention proposes an end-to-end robotic arm guidance and calibration method based on pose detection and compensation, which includes: Step 1: Set multiple vehicle body scanning positions on the production line chain and acquire the corresponding vehicle body point cloud data through the acquisition units set at each vehicle body scanning position; Step 2: Register the vehicle body point cloud data with the standard vehicle body point cloud and extract the vehicle body reference point position information; Step 3: Based on the vehicle body reference point position information, obtain the spatial transformation data of the vehicle body center point coordinates relative to the production line plate chain coordinate system; and obtain the position of the vehicle body center point in the robot arm base coordinate system, thus obtaining the position of the vehicle body center point in the robot arm base coordinate system. Step 4: Install an encoder on the chain drive shaft to obtain the chain speed and predict the position of the vehicle body center point in the robot arm's base coordinate system at the next moment; thus, predict the position change of the vehicle body in real time.
[0023] Based on the core steps, this application firstly: uses a vehicle pose calculation method based on four sets of 3D profilometers, and employs a point cloud matching algorithm to obtain the pose information of the actual vehicle body relative to a reference vehicle body; secondly: uses a high-precision encoder combined with the vehicle body pose calculation results to predict the pose coordinates of the vehicle body on the production line at the next moment (at which time the robotic arm can reach the target reference point). The robotic arm can actually measure (this application directly uses a 3D camera at the end of the robotic arm) the coordinates of the target reference point relative to the robotic arm's base coordinate system. Thus, with two sets of coordinate points—the actual target reference point and the predicted target reference point—and sufficient data collected based on the measurement results of the same vehicle body, an error model between the predicted and actual measured values can be established. A neural network is used for error compensation, and through training, a nonlinear compensation relationship is obtained, achieving error compensation from the actual measured point to the target point. This establishes an accurate transformation relationship between the 3D profilometer data and the robot's base coordinate system. This avoids the accumulation of errors from separate calibrations of the robot and the production line, and between the 3D profilometer and the production line, and is also easy to operate.
[0024] In a further embodiment, when detecting the vehicle body pose, it is necessary to acquire real-time vehicle body point cloud data and standard vehicle body point cloud data. Specifically, step 1 mentioned above includes: planning scanning positions based on the vehicle body's geometric features, i.e., setting multiple vehicle body scanning positions, and acquiring corresponding vehicle body point cloud data by setting acquisition units at each vehicle body scanning position. The vehicle body scanning positions are several scanning positions evenly distributed around the vehicle body, such as four positions on the left front, left rear, right front, and right rear of the vehicle body, representing a 360° key area of the vehicle body. The acquisition unit preferably uses four 3D profilometers to synchronously trigger scanning of the vehicle body to acquire vehicle body point cloud data. A 3D profilometer is a high-precision non-contact optical measurement device specifically designed for quickly acquiring the three-dimensional shape (contour, height, depth, etc.) of an object's surface. In vehicle body scanning applications, it uses technologies such as laser triangulation or structured light projection to convert the geometric shape of the vehicle body surface into dense point cloud data.
[0025] In a further embodiment, step 2 mentioned above includes: registering the vehicle body point cloud data with a standard vehicle body point cloud (providing a guiding coordinate system reference for the robotic arm), and extracting vehicle body reference points to obtain their coordinates in the production line chain coordinate system; that is, performing point cloud registration with the point cloud of the area corresponding to the standard vehicle body to further obtain multiple corresponding vehicle body reference point information; such as unifying the vehicle body point cloud data collected by four 3D contour machines to the same coordinate system (such as the production line chain coordinate system); specific steps include: Assume the point cloud data measured by the four 3D profilometers are represented as follows: This ensures that each 3D profiler corresponds to a point cloud of a specific area of the vehicle body; the standard vehicle body point cloud is represented as: ; Based on the ICP algorithm, the calibration matrix required for the point cloud registration transformation process from the 3D profiler to the corresponding area of the vehicle body is represented as follows: The calculation formula is as follows:
[0026] Where p represents a point in the point cloud data measured by the 3D profilometer, p ’ Represented as points on the initial standard vehicle body point cloud. i It is the corresponding number of the profiler. T is the spatial change data matrix of each sensor, namely the 3D profiler and the initial standard body. The initial standard body is rigidly connected to the plate chain coordinate system through the positioning fixture. Its standard point cloud serves as the reference for all dynamic body registration. That is, the initial body can be used as the reference for all other bodies relative to the production line plate chain. The registration operation is performed by solving the calibration matrix to obtain the position of the vehicle body reference point in the corresponding area. In the vehicle body scanning system, the point cloud acquired by the 3D profilometer is registered in real time with the standard vehicle body point cloud using a transformation matrix, and reference points are extracted; the formula for calculating the reference points is as follows: p i ref = T i sensor→std p i sensor ; Where, p i sensor This is represented as point cloud data measured by a 3D profilometer.
[0027] In a further embodiment, step 3 described above includes: Step 31: Based on the vehicle body reference point position information, obtain the spatial transformation data (translation and rotation) of the vehicle body center point coordinates relative to the production line chain coordinate system; specifically, this refers to obtaining the target position obtained through registration, i.e., the vehicle body reference point position information. Point cloud positions of each standard vehicle body The corresponding spatial transformation data, i.e., the rotation matrix, is obtained by solving the rigid body transformation model. The translation vector, the corresponding calculation formula is:
[0028] in: .
[0029] Step 32: Given that the robot arm base coordinate system and the production line chain coordinate system have a corresponding spatial transformation relationship, the spatial position of the vehicle body center point relative to the robot arm base coordinate system at the same moment can be obtained; specifically, based on the spatial transformation data of the vehicle body center point coordinates relative to the production line chain coordinate system, and based on the aforementioned relationship, the position of the vehicle body center point in the robot arm base coordinate system is obtained, and the corresponding calculation formula is as follows;
[0030] in, This represents the transformation relationship from the robot arm's base coordinate system to the production line's chain coordinate system. T base→line The corresponding rough value can be obtained by the relative reference position of the installed mechanism (e.g., by estimating through the installation structure model and layout diagram, or by directly measuring with measuring tools, such as laser rangefinders). Here, 'center' represents the coordinates of the vehicle's center point in the standard vehicle coordinate system. The aforementioned standard vehicle coordinate system is a local coordinate system fitted from a standard vehicle point cloud. Its establishment method includes: selecting the center points of the four tire mounting positions in the standard point cloud; establishing a right-handed coordinate system with the center point as the origin and the vehicle's front direction as the X-axis; that is, using the previously obtained coordinates of the four key points p... i std The solution can be simply understood as the position of the center point of the four tires. The center point of the vehicle can be obtained by finding the center line of the coordinates of the four key points.
[0031] Step 33, given p car base That is, p center base Then, the position of the vehicle's center point in the robot arm's base coordinate system can be obtained, and the corresponding formula is: ; Step 3 first calculates the spatial transformation of the vehicle body center point coordinates relative to the production line plate chain coordinate system, and then transforms it to the robot arm base coordinate system to obtain the position of the vehicle body center point in the robot arm base coordinate system.
[0032] In a further embodiment, step 4 above includes: installing an encoder on the chain drive shaft, using the encoder to acquire the chain speed, and using a kinematic model to predict the position of the vehicle's center point in the base coordinate system at the next moment; thereby predicting the vehicle's position change data in real time. The chain speed is obtained by acquiring the chain's linear velocity in real time using a high-precision encoder installed on the chain drive shaft. Then, by using the speed of the chain obtained by the encoder in this step, the position of the vehicle center point in the base coordinate system at the next moment can be predicted. Based on this, the position of a certain measurement point in the vehicle center point coordinate system relative to the base coordinate system can be located for subsequent visual guidance.
[0033] Preferably, the chain speed is obtained using an encoder. The formula for predicting the position of the vehicle's center point in the base coordinate system at the next moment is:
[0034] Preferably, the process of locating the position of a measurement point in the vehicle center coordinate system relative to the base coordinate system and then performing subsequent visual guidance also includes: Combining real-time data from visual measurements, through an error model Establish the following formula:
[0035] Where, p predicted =p center base (t+Δt) + p, where p is the position relative to the robot's base coordinate system. measured This involves transforming the 3D position information obtained from the actual measurement by the 3D camera mounted on the end effector of the robotic arm into the measured position in the base coordinate system; while the 3D position information obtained by the 3D camera is relative to the camera coordinate system, and the camera coordinate system is used to obtain the TCP-based position p through hand-eye calibration. target TCP p target TCP The coordinates can be unified to the base coordinate system through TCP-to-base coordinate system transformation; that is, p base target Can be regarded as p measured .
[0036] If the relative TCP position of the boot target point is Then its position in the base coordinate system is:
[0037] in, The transformation relationship from the end effector TCP of the robotic arm to the base coordinate system (can be directly derived from the kinematic model of the robotic arm itself), that is, it can be directly read from the state parameters of the robotic arm, which are generally in the form of x, y, z, rx, ry, rz, and can be converted into the form of transformation matrix.
[0038] Furthermore, since the relative position of the target point to be guided by the same model is fixed relative to the TCP of the robotic arm, an error model corresponding to the kinematic model can be established according to the above process. If the 3D position of the target point relative to the TCP can be specifically measured, the corresponding parameters can be identified. Using neural network methods or parameter identification schemes, error compensation can be performed directly or 3D profilometer data can be obtained, ultimately leading to the transformation relationship in the robot's base coordinate system.
[0039] Preferably, to achieve high-precision position guidance, neural networks or parameter identification methods can be used to learn the error model:
[0040] Nonlinear compensation relationships are obtained through training. This enables error compensation between the actual measured point and the target point, thereby establishing an accurate transformation relationship between the profilometer data and the robot's base coordinate system.
[0041] This solution achieves precise position calibration and guidance of the robotic arm through multi-sensor data fusion, spatial transformation, and error modeling. The solution essentially combines model building and learning compensation capabilities.
[0042] Based on the same inventive concept, the present invention also provides an end-to-end robotic arm guidance and calibration system based on pose detection and compensation, comprising: The data acquisition unit is used to set multiple vehicle body scanning positions and acquire the corresponding vehicle body point cloud data by setting the acquisition unit at each vehicle body scanning position; The data registration unit is used to register the vehicle body point cloud data with the standard vehicle body point cloud to obtain the position information of multiple vehicle body reference points; The data calculation unit is used to obtain the spatial transformation data of the coordinates of the center point of the vehicle body relative to the coordinate system of the production line plate chain based on the position information of the vehicle body reference point; and to obtain the position of the center point of the vehicle body in the base coordinate system, that is, to obtain the position of the center point of the vehicle body in the base coordinate system of the robotic arm. The data prediction unit is used to install on the chain drive shaft and use an encoder to obtain the chain speed, predict the position of the vehicle body center point in the base coordinate system at the next moment, and thus predict the position change of the vehicle body in real time.
[0043] This invention can be applied to scenarios involving precise manipulation of car bodies by robotic arms in fields such as automobile manufacturing and repair, including tasks like welding, painting, and assembly. In these scenarios, the robotic arm needs to accurately position and position the car body to complete the corresponding tasks. The method of this invention can improve the operational accuracy and efficiency of the robotic arm, reduce calibration costs, minimize human intervention, and improve production quality and safety.
[0044] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.
Claims
1. An end-to-end robotic arm guidance and calibration method based on pose detection and compensation, characterized in that, It includes: Step 1: Set multiple vehicle body scanning positions on the production line chain and acquire the corresponding vehicle body point cloud data through the acquisition units set at each vehicle body scanning position; Step 2: Register the vehicle body point cloud data with the standard vehicle body point cloud to obtain the position information of multiple vehicle body reference points; Step 3: Based on the vehicle body reference point position information, obtain the spatial transformation data of the vehicle body center point coordinates relative to the production line plate chain coordinate system; and obtain the position of the vehicle body center point in the robot arm base coordinate system to obtain the position of the vehicle body center point in the robot arm base coordinate system. Step 4: Install an encoder on the chain drive shaft to obtain the chain speed and predict the position of the vehicle body center point in the robot arm's base coordinate system at the next moment; thus, predict the position change of the vehicle body in real time.
2. The end-to-end robotic arm guidance and calibration method based on pose detection and compensation as described in claim 1, characterized in that, In step 1, the vehicle body scanning positions are a number of scanning positions evenly distributed around the vehicle body; the acquisition unit uses a 3D profilometer to synchronously trigger a scan of the vehicle body to obtain the vehicle body point cloud data.
3. The end-to-end robotic arm guidance and calibration method based on pose detection and compensation as described in claim 1, characterized in that, Step 2 includes: Assume the point cloud data measured by four 3D profilometers are represented as follows: This ensures that each 3D profiler corresponds to a point cloud of a specific area of the vehicle body; the standard vehicle body point cloud is represented as... ; The calibration matrix for the point cloud registration transformation process from the 3D profiler to the corresponding area of the vehicle body is represented as follows: ;in, ， Where p represents a point in the measurement point cloud data corresponding to the 3D profilometer, p ’ Represented as points on the initial standard vehicle body point cloud; The registration operation is performed by solving the calibration matrix to obtain the position of the vehicle body reference point in the corresponding area. In the vehicle body scanning system, a transformation matrix is used to register the point cloud collected by the 3D profilometer with the standard vehicle body point cloud in real time and extract the position information of the vehicle body reference points; among which, p i ref = T i sensor→std p i sensor 。 4. The end-to-end robotic arm guidance and calibration method based on pose detection and compensation as described in claim 1, characterized in that, Step 3 mentioned above includes: Step 31: Based on the vehicle body reference point position information, obtain the spatial transformation data of the vehicle body center point coordinates relative to the production line chain coordinate system; specifically, this refers to obtaining the spatial transformation data of the vehicle body center point coordinates relative to the production line chain coordinate system based on the vehicle body reference point position information. Point cloud positions of each standard vehicle body The corresponding spatial transformation data is obtained by solving the rigid body transformation model, and the corresponding solution formula is as follows: ， in: ; Step 32: Obtain the spatial position of the vehicle's center point relative to the robot arm's base coordinate system at the same moment; the corresponding calculation formula is as follows; ， in, This represents the transformation relationship from the robot arm's base coordinate system to the production line's chain coordinate system. These are the coordinates of the vehicle's center point in the standard vehicle coordinate system, where 'center' represents the center point in the standard vehicle coordinate system. Step 33: Obtain the position of the vehicle body center point in the robot arm's base coordinate system. The corresponding formula is: 。 5. The end-to-end robotic arm guidance and calibration method based on pose detection and compensation as described in claim 4, characterized in that, Step 4 above uses an encoder to obtain the speed of the chain movement. The formula for predicting the position of the vehicle's center point in the base coordinate system at the next moment is: 。 6. A calibration system for performing the method as described in any one of claims 1-5, characterized in that, The data acquisition unit is used to set multiple body scanning positions on the production line chain and acquire the corresponding body point cloud data through the acquisition units set at each body scanning position. The data registration unit is used to register the vehicle body point cloud data with the standard vehicle body point cloud to obtain the position information of multiple vehicle body reference points; The data calculation unit is used to obtain the spatial transformation data of the coordinates of the center point of the vehicle body relative to the production line plate chain coordinate system based on the vehicle body reference point position information; and to obtain the position of the center point of the vehicle body in the robot arm base coordinate system, that is, to obtain the position of the center point of the vehicle body in the robot arm base coordinate system. The data prediction unit is used to install on the chain drive shaft and use an encoder to obtain the chain speed, and predict the position of the vehicle body center point in the robot arm base coordinate system at the next moment; thus, it can predict the position change of the vehicle body in real time.