Robot tool coordinate system automatic positioning method based on point cloud and model registration

Abstract

The invention discloses a robot tool coordinate system automatic positioning method based on point cloud and model registration, and relates to the field of robot automatic grinding and polishing. The method comprises the following steps: carrying out eye outside-hand calibration on a robot and a scanner; an original point of a coordinate system of the tool design model is established at the same position of a machining TCP point of the standard physical tool, the coordinate system directions of the standard physical tool and the tool design model are set to be consistent, teaching calibration of the tool coordinate system under the robot is carried out, and a teaching tool coordinate system matrix is obtained; summarizing the hand-eye calibration matrix and the teaching tool coordinate system matrix, obtaining standard physical tool point cloud data and corresponding robot poses, and calculating to obtain a standard physical tool model under a tool design model coordinate system and a corresponding standard physical-design model conversion matrix; and obtaining point cloud data of a to-be-measured physical tool and a corresponding robot pose, calculating to obtain a tool coordinate system of visual calibration under the robot coordinate system, and realizing positioning of the tool.

Description

Technical Field

[0001] The present invention relates to the field of robot automated grinding and polishing, and in particular to a method for automatic positioning of a robot tool coordinate system based on point cloud and model registration. Background Art

[0002] Currently, the field of robotic automated grinding and polishing still generally relies on manual teaching to point the tool coordinate system. This method has significant limitations: First, the manual operation mode makes the calibration process time-consuming and inconsistent results difficult to ensure; second, the need for repeated calibration after tool replacement or wear results in continuous resource consumption; more importantly, differences in operator skills can introduce difficult-to-control random errors. These systematic flaws not only restrict production efficiency improvements but also become a key bottleneck hindering the industry's intelligent transformation. Therefore, the development of highly robust automated calibration technology has important engineering value. Summary of the Invention

[0003] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a method for automatic positioning of a robot tool coordinate system based on point cloud and model registration, combining theoretical calculation with automated control to achieve automation of positioning and correction of the robot tool coordinate system.

[0004] The technical solution of the present invention is: A method for automatic positioning of a robot tool coordinate system based on point cloud and model registration, the method comprising the following steps: Step 1: Robot in the grinding and polishing system and scanner Perform eye calibration outside the hand to obtain the image used to represent the robot The center of the base and the scanner The hand-eye calibration matrix of the positional relationship between the camera coordinate system ; Step 2: Establish the coordinate system origin of the tool design model at the same position as the machining TCP point of the standard physical tool, and set the coordinate system direction of the standard physical tool to be consistent with the coordinate system direction of the tool design model. This is used to perform teaching calibration of the tool coordinate system in the robot coordinate system, and obtain the teaching tool coordinate system matrix of the standard physical tool in the robot coordinate system; Step 3: Summarize the hand-eye calibration matrix obtained in step 1 And the teaching tool coordinate system matrix obtained by step 2 , and obtain the point cloud data of the standard physical tool and the corresponding robot posture, calculate the standard physical tool model in the tool design model coordinate system and the corresponding standard physical-design model conversion matrix; Step 4: Based on the standard physical tool in the tool design model coordinate system obtained in step 3 and the corresponding standard physical-design model conversion matrix, and obtaining the point cloud data of the physical tool to be measured and the corresponding robot pose, calculate the tool coordinate system for visual calibration in the robot coordinate system to achieve tool positioning.

[0005] Furthermore, according to the automatic positioning method of the robot tool coordinate system, the robot in the grinding and polishing system described in step 1 and scanner The process of eye-outside-hand calibration is as follows: Step 1.1: Fix the prepared calibration ball on the robot in a certain posture The end of the flange, the scanner Fixed to robot A certain position outside to ensure that the mobile robot passes Enable Scanner In a convenient position for scanning the calibration sphere; Step 1.2: Move the Robot , so that the calibration ball is in the scanner When the scanner is within the field of view, Scan the surface of the calibration sphere to collect point cloud data , and record the robot In its base coordinate system and tool coordinate system The coordinate value of ; Step 1.3: Repeat step 1.2 until you have four groups of robots. Posture to And the scanner in the corresponding position Collected point cloud ; Step 1.4: Repeat step 1.3 until you have six groups of robots. Posture to And the scanner in the corresponding position Collected point cloud ; Step 1.5: Scanner Collected point cloud Perform filtering and cropping. If the number of points in the point cloud exceeds the set value, additional downsampling is performed to obtain a post-processed point cloud containing only the calibration sphere point cloud. ; Step 1.6: Using the robot's position coordinates to and point cloud Calculate the hand-eye calibration matrix of Quantity; and use Component, robot position coordinate value to and point cloud Calculate the hand-eye calibration matrix of Component, that is, the hand-eye calibration matrix ; Step 1.7: Use the obtained hand-eye calibration matrix , robot position coordinate value to and point cloud Calculate 10 in the scanner Hemispherical point cloud in coordinate system is merged into robot Complete spherical point cloud in coordinate system for:

[0006] Step 1.8: To the robot Complete spherical point cloud in coordinate system Do spherical fitting and get the fitting radius RMS error ; According to the measurement radius of the calibration ball in the given calibration ball detection report , get the fitting radius error for:

[0007] Step 1.9: Repeat step 1.8 several times to Complete spherical point cloud in coordinate system Do spherical fitting and take the average value to get the final average fitting radius error and the average root mean square error ; Step 1.10: Judgement and Do they all meet the accuracy requirements required in the use environment? If not, repeat steps 1.1 to 1.7 until the requirements are met to obtain the final robot. and scanner The calibration matrix of the eye outside the hand is .

[0008] Furthermore, according to the robot tool coordinate system automatic positioning method, step 1.3 obtains four groups of robots Posture to And the scanner in the corresponding position Collected point cloud , the following requirements must be met: ①Robot These four sets of postures 、 、 The quantities vary, and 、 、 The quantities are the same; ②Robot These four sets of postures 、 、 The weight difference between the two weights should be as large as possible and not less than 5mm; ③Scanner Collected point cloud It should be ensured that: each point cloud contains half a complete and independent spherical point cloud without interference, that is, it can be finally cropped into an independent hemispherical point cloud; the spherical point cloud is uniform and smooth, without large noise data.

[0009] Furthermore, according to the robot tool coordinate system automatic positioning method, step 1.4 obtains six groups of robots Posture to And the scanner in the corresponding position Collected point cloud , meeting the following requirements: ①Robot These six sets of postures 、 、 、 、 、 Quantities vary; ②Robot These six sets of postures 、 、 The weight difference between the two weights should be as large as possible and not less than 5mm; the six groups of robot postures 、 、 The difference between the weights should be as large as possible and not less than 5°; ③Scanner Collected point cloud It should be ensured that: each point cloud contains half a complete and independent spherical point cloud without interference, that is, it can be finally cropped into an independent hemispherical point cloud; the spherical point cloud is uniform and smooth, without large noise data.

[0010] Furthermore, according to the robot tool coordinate system automatic positioning method, step 3 further includes the following steps: Step 3.1: Obtain the point cloud data of the standard physical tool and the corresponding robot pose during the actual processing process, and convert the point cloud data of the standard physical tool from the scanner coordinate system to the robot coordinate system; Step 3.2: Obtain the standard physical tool model in the tool design model coordinate system and the corresponding standard physical-design model conversion matrix ; Step 3.2.1: N The tool point cloud obtained in the robot coordinate system during the scanning process is filtered, cropped and merged to obtain the point cloud of the standard physical tool in the robot coordinate system:

[0011] in, Represents the original point cloud The result after filtering and cropping is that only valid matching point clouds are retained and noise and external interferences are deleted; Step 3.2.2: Obtain the tool design model in point cloud form ; Combined with the teaching tool coordinate system matrix calibrated in the robot coordinate system The calculated point cloud of the standard physical tool in the tool design model coordinate system is:

[0012] Step 3.2.3: Apply the ICP algorithm to calculate the transfer matrix of the standard physical tool point cloud offset to the tool design model in the tool design model coordinate system:

[0013] The standard physical tool model in the form of point cloud in the tool design model coordinate system is calculated as follows:

[0014] Step 3.2.4: According to the calibration of the teaching tool coordinate system matrix in the robot coordinate system Transfer matrix of the standard physical tool point cloud offset to the tool design model in the tool design model coordinate system , calculate the standard physical-design model conversion matrix as: .

[0015] Furthermore, according to the robot tool coordinate system automatic positioning method, the process of converting the standard physical tool point cloud data from the scanner coordinate system to the robot coordinate system in step 3.1 is as follows: Step 3.1.1: The robot moves to a certain position , the scanner collects the point cloud data of the standard physical tool at that position, ensuring that the point cloud data can cover a certain plane or curved surface of the standard physical tool; wherein each point of the point cloud data is , then there is a corresponding point cloud ; Step 3.1.2: Repeat step 3.1.1 until you get Point cloud data The corresponding robot position , and record the robot pose matrix at the corresponding position; ensure this The point cloud at each position can cover the core features of standard physical tools; Step 3.1.3: Summarize the robot's position at the point during the scan obtained in step 3.1.2 The pose matrix Point cloud of standard physical tool in the corresponding scanner coordinate system , calculated robot The position transfer matrix of scanner C during scanning for:

[0016] Further calculation shows that during the scanning process, the robot The standard physical tool point cloud below is: .

[0017] Furthermore, according to the robot tool coordinate system automatic positioning method, step 4 further includes the following steps: Step 4.1: Obtain the point cloud data of the physical tool to be measured and the corresponding robot pose during the actual processing process, and convert the point cloud data of the physical tool to be measured from the scanner coordinate system to the robot coordinate system; Step 4.1.1: Robot moves to position , the scanner collects the point cloud data of the physical tool to be measured at this position; each point of the point cloud data is , then there is a corresponding point cloud ; Step 4.1.2: Repeat step 4.1.1 until you get Point cloud data The corresponding robot position ; Step 4.1.3: Summarize the robot's position at the point during the scan obtained in step 4.1.2 The pose matrix and the point cloud of the physical tool to be measured in the corresponding scanner coordinate system , the point cloud of the physical tool to be measured in the robot coordinate system during the scanning process is calculated as:

[0018] Step 4.2: Obtain the tool coordinate system for visual calibration in the robot coordinate system , realize the positioning of the tool; Step 4.2.1: N The tool point cloud obtained in the robot coordinate system during the scanning process is filtered, cropped and merged to obtain the point cloud of the standard physical tool under the robot:

[0019] in, Represents the original point cloud The result after filtering and cropping is that only valid matching point clouds are retained and noise and external interferences are deleted; Step 4.2.2: Obtain the standard physical tool model in the tool design model coordinate system , combined with the standard physical-design model conversion matrix The calculated point cloud of the standard physical tool in the robot coordinate system is:

[0020] Step 4.2.3: Apply the ICP algorithm to calculate the transfer matrix from the point cloud of the standard physical tool model in the robot coordinate system to the point cloud of the physical tool to be measured:

[0021] Step 4.2.4: Transfer matrix from the point cloud of the standard physical tool model in the robot coordinate system to the point cloud of the physical tool to be measured Standard physical object-design model conversion matrix , calculate the tool coordinate system matrix of the visual calibration in the robot coordinate system: .

[0022] Compared with the prior art, the present invention has the following beneficial effects: (1) After the preparation stage is completed, the calibration in the subsequent application stage only requires the collection of point clouds and matrix calculations, which can reduce the manual calibration process from several minutes to tens of minutes to several seconds to tens of seconds, greatly improving efficiency.

[0023] (2) The positioning process of the tool in the present invention is not affected by the operator's operating errors. Its accuracy is only limited by the point cloud accuracy of the scanner and the positioning accuracy of the robot. The results have stronger consistency, which provides a guarantee for the stability of the automated process.

[0024] (3) Compared with the manual teaching method, the positioning results of the present invention for unused standard tools have been improved by the model-based guidance. The high-precision positioning of used tools or tools with certain deformation cannot be achieved by the manual teaching method. Therefore, the present invention has more advanced and more referenceable features. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 Schematic diagram of the structure of the grinding and polishing system of this embodiment; Figure 2 This is a flow chart of the method for automatic positioning of a robot tool coordinate system based on point cloud and model registration in this embodiment; The accompanying drawings illustrate: 1—Robot, 2—Scanner, 3—Robotic tool changer, 4—Grinding tool. DETAILED DESCRIPTION

[0026] To facilitate understanding of the present application, the present application will be described more comprehensively below with reference to the relevant drawings.

[0027] This embodiment is based on the robot tool coordinate system automatic positioning method based on point cloud and model registration. Figure 1 The polishing system shown in the figure is implemented. The polishing system includes a robot 1 carrying a robot tool changer 3 and a polishing tool 4, and a scanner 2 for positioning the polishing tool 4. The robot 1 places the polishing tool 4 within the field of view of the scanner 2 by teaching it several positions. Through photography and matrix transformation, the tool coordinate system of the polishing tool 4 within the robot system is calculated and used to complete the workpiece polishing.

[0028] For the convenience of formula description, only Figure 2 In the calculation process, the identification of robot 1 is defined as , define the identifier of scanner 2 as .

[0029] Figure 2 This is a flow chart of the method for automatic positioning of the robot tool coordinate system based on point cloud and model registration in this embodiment. Figure 2 As shown, the method for automatic positioning of the robot tool coordinate system based on point cloud and model registration includes the following steps: Step 1: Targeting the Robot in the Grinding and Polishing System and scanner Perform eye-outside-hand calibration to obtain the eye-outside-hand calibration result matrix.

[0030] This step is the preparation phase 1, the robot in the grinding and polishing system and scanner Perform eye-outside-hand calibration to obtain the eye-outside-hand calibration result matrix , that is, the hand-eye calibration matrix, which represents the robot The center of the base and the scanner The positional relationship between the camera coordinate systems.

[0031] This embodiment performs robot and scanner The process of eye calibration outside the hand is as follows: Step 1.1: Fix the prepared calibration ball on the robot in a certain posture The end of the flange, the scanner Fixed to robot A certain position outside to ensure that the mobile robot passes Enable Scanner In a convenient position for scanning the calibration sphere; In this embodiment, a calibration ball with an accuracy test report is prepared, the detection radius of the calibration ball is obtained, and a calibration ball can be fixed on the robot. Fix the calibration ball to the robot in a certain posture. The flange end ensures that it is in contact with the robot during the subsequent point cloud collection process. Tool coordinate system The relative position relationship does not change and will not be affected by occlusion, interference, etc.; mobile robot Enable Scanner In a convenient position for scanning the calibration sphere.

[0032] In addition, define the robot and scanner The calibration result matrix of the eye outside the hand and its Cartesian expression is:

[0033] Step 1.2: Move the Robot , so that the calibration ball is in the scanner When the scanner is within the field of view, Scan the surface of the calibration sphere to collect point cloud data , and record the robot In its base coordinate system and tool coordinate system The coordinate value of .

[0034] Step 1.3: Repeat step 1.2 until you have four groups of robots. Posture to And the scanner in the corresponding position Collected point cloud , and ensure that the following requirements are met: ①These four groups of robots Postural 、 、 The quantities vary, and 、 、 The quantities are the same, that is

[0035] ②These four groups of robots Postural 、 、 The weight difference between the two weights should be as large as possible and not less than 5mm, that is,

[0036] ③Scanner Collected point cloud It should be ensured that: each point cloud contains half a complete and independent spherical point cloud without interference, that is, it can be finally cropped into an independent hemispherical point cloud; the spherical point cloud is uniform and smooth, without large noise data.

[0037] Step 1.4: Repeat step 1.3 until you have six groups of robots. Posture to And the scanner in the corresponding position Collected point cloud , and ensure that the following requirements are met: ① These six groups of robot postures 、 、 、 、 、 The quantities vary, i.e.

[0038] ② These six groups of robot postures 、 、 The weight difference between the two weights should be as large as possible and not less than 5mm; the six groups of robot postures 、 、 The difference between the two components should be as large as possible and not less than 5°, that is,

[0039] ③Scanner Collected point cloud It should be ensured that: each point cloud contains half a complete and independent spherical point cloud without interference, that is, it can be finally cropped into an independent hemispherical point cloud; the spherical point cloud is uniform and smooth, without large noise data.

[0040] Step 1.5: Scanner Collected point cloud Perform filtering and cropping. If the number of points in the point cloud exceeds the set value, additional downsampling is performed to obtain a post-processed point cloud containing only the calibration sphere point cloud. .

[0041] Step 1.6: Using the robot's position coordinates to and point cloud Calculate the hand-eye calibration matrix of Quantity; and use Component, robot position coordinate value to and point cloud Calculate the hand-eye calibration matrix of Component, that is, the hand-eye calibration matrix .

[0042] Step 1.7: Use the obtained hand-eye calibration matrix , robot position coordinate value to and point cloud Calculate 10 in the scanner Hemispherical point cloud in coordinate system is merged into robot Complete spherical point cloud in coordinate system for:

[0043] Step 1.8: To the robot Complete spherical point cloud in coordinate system Do spherical fitting and get the fitting radius RMS error ; According to the measurement radius of the calibration ball in the given calibration ball detection report , get the fitting radius error for:

[0044] Step 1.9: To eliminate accidental errors, repeat step 1.8 several times to adjust the robot. Complete spherical point cloud in coordinate system Take the average value of the ball fitting and ensure that the evaluation unit is mm to obtain the final average fitting radius error and the average root mean square error ; Step 1.10: Judgement and Whether they meet the accuracy requirements required in the use environment, qualified calibration results should meet the accuracy requirements required in the use environment. This example has the following requirements:

[0045] If not, repeat steps 1.1 to 1.7 until the requirements are met and the final robot is obtained. and scanner The calibration matrix of the eye outside the hand is .

[0046] Step 2: Perform preparation phase 2. Establish the coordinate system origin of the tool design model at the same location as the machining TCP point of the standard physical tool, and set the coordinate system direction of the standard physical tool to be consistent with the coordinate system direction of the tool design model. This is used to perform teaching calibration of the tool coordinate system in the robot coordinate system, and obtain the teaching tool coordinate system matrix of the standard physical tool in the robot coordinate system. Step 2.1: Reconstruct the coordinate system of the tool design model, set the origin to coincide with the TCP point of the standard physical tool, and set the direction to be consistent with the coordinate system direction required by the standard physical tool; Step 2.2: Set the tool coordinate system on the robot The calibration process is carried out according to the standard process specified by the robot used, and the calibration results are obtained by the tool on the robot. The coordinate system matrix of the teaching tool under .

[0047] Step 3: Perform preparation phase 3 and summarize the hand-eye calibration matrix obtained in step 1 And the teaching tool coordinate system matrix obtained by step 2 , and obtain the point cloud data of the standard physical tool and the corresponding robot posture, calculate the standard physical tool model in the tool design model coordinate system and the corresponding standard physical-design model conversion matrix; Step 3.1: Obtain the point cloud data of the standard physical tool and the corresponding robot pose during the actual processing process, and convert the point cloud data of the standard physical tool from the scanner coordinate system to the robot coordinate system; Step 3.1.1: The robot moves to a certain position , the scanner collects the point cloud data of the standard physical tool at that position, ensuring that the point cloud data can cover a certain plane or curved surface of the standard physical tool; wherein each point of the point cloud data is , then there is a corresponding point cloud ; Step 3.1.2: Repeat step 3.1.1 until you get Point cloud data The corresponding robot position , and record the robot pose matrix at the corresponding position; ensure this The point cloud at each position can cover the core features of standard physical tools; Step 3.1.3: Summarize the robot's position at the point during the scan obtained in step 3.1.2 The pose matrix Point cloud of standard physical tool in the corresponding scanner coordinate system , calculated robot The position transfer matrix of scanner C during scanning for:

[0048] Further calculation shows that during the scanning process, the robot The standard physical tool point cloud below is:

[0049] Step 3.2: Obtain the standard physical tool model in the tool design model coordinate system and the corresponding standard physical-design model conversion matrix ; Step 3.2.1: N The tool point cloud obtained in the robot coordinate system during the scanning process is filtered, cropped and merged to obtain the point cloud of the standard physical tool in the robot coordinate system:

[0050] in, Represents the original point cloud The result after filtering and cropping is that only valid matching point clouds are retained and noise and external interferences are deleted; Step 3.2.2: Obtain the tool design model in point cloud form ; Combined with the teaching tool coordinate system matrix calibrated in the robot coordinate system The calculated point cloud of the standard physical tool in the tool design model coordinate system is:

[0051] Step 3.2.3: Apply the ICP algorithm to calculate the transfer matrix of the standard physical tool point cloud offset to the tool design model in the tool design model coordinate system:

[0052] The standard physical tool model in the form of point cloud in the tool design model coordinate system is calculated as follows:

[0053] Step 3.2.4: According to the calibration of the teaching tool coordinate system matrix in the robot coordinate system Transfer matrix of the standard physical tool point cloud offset to the tool design model in the tool design model coordinate system , calculate the standard physical-design model conversion matrix as:

[0054] Step 4: Carry out the formal application part, that is, the part that needs to be run repeatedly in actual applications. Based on the standard physical tool in the tool design model coordinate system obtained in step 3 and the corresponding standard physical-design model conversion matrix, the point cloud data of the physical tool to be measured and the corresponding robot pose are obtained, and the tool coordinate system of the visual calibration in the robot coordinate system is calculated to realize the positioning of the tool.

[0055] Step 4.1: Obtain the point cloud data of the physical tool to be measured and the corresponding robot pose during the actual processing process, and convert the point cloud data of the physical tool to be measured from the scanner coordinate system to the robot coordinate system; Step 4.1.1: Robot moves to position , the scanner collects the point cloud data of the physical tool to be measured at this position; each point of the point cloud data is , then there is a corresponding point cloud ; Step 4.1.2: Repeat step 4.1.1 until you get Point cloud data The corresponding robot position ; Step 4.1.3: Summarize the robot's position at the point during the scan obtained in step 4.1.2 The pose matrix and the point cloud of the physical tool to be measured in the corresponding scanner coordinate system , the point cloud of the physical tool to be measured in the robot coordinate system during the scanning process is calculated as:

[0056] Step 4.2: Obtain the tool coordinate system for visual calibration in the robot coordinate system , realize the positioning of the tool; Step 4.2.1: N The tool point cloud obtained in the robot coordinate system during the scanning process is filtered, cropped and merged to obtain the point cloud of the standard physical tool under the robot:

[0057] in, Represents the original point cloud The result after filtering and cropping is that only valid matching point clouds are retained and noise and external interferences are deleted; Step 4.2.2: Obtain the standard physical tool model in the tool design model coordinate system , combined with the standard physical-design model conversion matrix The calculated point cloud of the standard physical tool in the robot coordinate system is:

[0058] Step 4.2.3: Apply the ICP algorithm to calculate the transfer matrix from the point cloud of the standard physical tool model in the robot coordinate system to the point cloud of the physical tool to be measured:

[0059] Step 4.2.4: Transfer matrix from the point cloud of the standard physical tool model in the robot coordinate system to the point cloud of the physical tool to be measured Standard physical object-design model conversion matrix , calculate the tool coordinate system matrix of the visual calibration in the robot coordinate system:

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the present invention.

Claims

1. A method for automatic positioning of robot tool coordinate system based on point cloud and model registration, characterized in that: The method comprises the following steps: Step 1: Robot in the grinding and polishing system and scanner Perform eye calibration outside the hand to obtain the image used to represent the robot The center of the base and the scanner The hand-eye calibration matrix of the positional relationship between the camera coordinate system ; Step 2: Establish the coordinate system origin of the tool design model at the same position of the processing TCP point of the standard physical tool, and set the coordinate system direction of the standard physical tool to be consistent with the coordinate system direction of the tool design model. In this way, the tool coordinate system is calibrated in the robot coordinate system to obtain the teaching tool coordinate system matrix of the standard physical tool in the robot coordinate system. ; Step 3: Summary and , and obtain the standard physical tool point cloud data and the corresponding robot pose, calculate the standard physical tool model in the tool design model coordinate system and the corresponding standard physical-design model conversion matrix; Step 4: Based on the standard physical tool in the tool design model coordinate system obtained in step 3 and the corresponding standard physical-design model conversion matrix, and obtaining the point cloud data of the physical tool to be measured and the corresponding robot pose, calculate the tool coordinate system for visual calibration in the robot coordinate system to achieve tool positioning.

2. The method for automatic positioning of a robot tool coordinate system according to claim 1, wherein: As described in step 1, the robot in the grinding and polishing system and scanner The process of eye-outside-hand calibration is as follows: Step 1.1: Fix the prepared calibration ball on the robot in a certain posture The end of the flange, the scanner Fixed to robot A certain position outside to ensure that the mobile robot passes Enable Scanner In a convenient position for scanning the calibration sphere; Step 1.2: Move the Robot , so that the calibration ball is in the scanner When the scanner is within the field of view, Scan the surface of the calibration sphere to collect point cloud data , and record the robot In its base coordinate system and tool coordinate system The coordinate value of ; Step 1.3: Repeat step 1.2 until you have four groups of robots. Posture to And the scanner in the corresponding position Collected point cloud ; Step 1.4: Repeat step 1.3 until you have six groups of robots. Posture to And the scanner in the corresponding position Collected point cloud ; Step 1.5: Scanner Collected point cloud Perform filtering and cropping. If the number of points in the point cloud exceeds the set value, additional downsampling is performed to obtain a post-processed point cloud containing only the calibration sphere point cloud. ; Step 1.6: Using the robot's position coordinates to and point cloud Calculate the hand-eye calibration matrix of Quantity; and use Component, robot position coordinate value to and point cloud Calculate the hand-eye calibration matrix of Component, that is, the hand-eye calibration matrix ; Step 1.7: Use the obtained hand-eye calibration matrix , robot position coordinate value to and point cloud Calculate 10 in the scanner Hemispherical point cloud in coordinate system is merged into robot Complete spherical point cloud in coordinate system for: ； Step 1.8: To the robot Complete spherical point cloud in coordinate system Do spherical fitting and get the fitting radius RMS error ; According to the measurement radius of the calibration ball in the given calibration ball detection report , get the fitting radius error for: ； Step 1.9: Repeat step 1.8 several times to Complete spherical point cloud in coordinate system Do spherical fitting and take the average value to get the final average fitting radius error and the average root mean square error ; Step 1.10: Judgement and Do they all meet the accuracy requirements required in the use environment? If not, repeat steps 1.1 to 1.7 until the requirements are met to obtain the final robot. and scanner The calibration matrix of the eye outside the hand is .

3. The method for automatic positioning of a robot tool coordinate system according to claim 2, wherein: Obtain four groups of robots as described in step 1.3 Posture to And the scanner in the corresponding position Collected point cloud , the following requirements must be met: ①Robot These four sets of postures 、 、 The quantities vary, and 、 、 The quantities are the same; ②Robot These four sets of postures 、 、 The weight difference between the two weights should be as large as possible and not less than 5mm; ③Scanner Collected point cloud It should be ensured that: each point cloud contains half a complete and independent spherical point cloud without interference, that is, it can be finally cropped into an independent hemispherical point cloud; the spherical point cloud is uniform and smooth, without large noise data.

4. The method for automatic positioning of a robot tool coordinate system according to claim 2, wherein: Obtain six groups of robots as described in step 1.4 Posture to And the scanner in the corresponding position Collected point cloud , meeting the following requirements: ①Robot These six sets of postures 、 、 、 、 、 Quantities vary; ②Robot These six sets of postures 、 、 The weight difference between the two weights should be as large as possible and not less than 5mm; the six groups of robot postures 、 、 The difference between the weights should be as large as possible and not less than 5°; ③Scanner Collected point cloud It should be ensured that: each point cloud contains half a complete and independent spherical point cloud without interference, that is, it can be finally cropped into an independent hemispherical point cloud; the spherical point cloud is uniform and smooth, without large noise data.

5. The method for automatic positioning of a robot tool coordinate system according to claim 1, wherein: The step 3 further comprises the following steps: Step 3.1: Obtain the point cloud data of the standard physical tool and the corresponding robot pose during the actual processing process, and convert the point cloud data of the standard physical tool from the scanner coordinate system to the robot coordinate system; Step 3.2: Obtain the standard physical tool model in the tool design model coordinate system and the corresponding standard physical-design model conversion matrix ; Step 3.2.1: N The tool point cloud obtained in the robot coordinate system during the scanning process is filtered, cropped and merged to obtain the point cloud of the standard physical tool in the robot coordinate system: ； in, Represents the original point cloud The result after filtering and cropping is that only valid matching point clouds are retained and noise and external interferences are deleted; Step 3.2.2: Obtain the tool design model in point cloud form ; Combined with the teaching tool coordinate system matrix calibrated in the robot coordinate system The calculated point cloud of the standard physical tool in the tool design model coordinate system is: ； Step 3.2.3: Apply the ICP algorithm to calculate the transfer matrix of the standard physical tool point cloud offset to the tool design model in the tool design model coordinate system: ； The standard physical tool model in the form of point cloud in the tool design model coordinate system is calculated as follows: ； Step 3.2.4: According to the calibration of the teaching tool coordinate system matrix in the robot coordinate system Transfer matrix of the standard physical tool point cloud offset to the tool design model in the tool design model coordinate system , calculate the standard physical-design model conversion matrix as: 。 6. The method for automatic positioning of a robot tool coordinate system according to claim 5, characterized in that: Step 3.1 converts the point cloud data of the standard physical tool from the scanner coordinate system to the robot coordinate system. The process is as follows: Step 3.1.1: The robot moves to a certain position , the scanner collects the point cloud data of the standard physical tool at that position, ensuring that the point cloud data can cover a certain plane or curved surface of the standard physical tool; wherein each point of the point cloud data is , then there is a corresponding point cloud ; Step 3.1.2: Repeat step 3.1.1 until you get Point cloud data The corresponding robot position , and record the robot pose matrix at the corresponding position; ensure this The point cloud at each position can cover the core features of standard physical tools; Step 3.1.3: Summarize the robot's position at the point during the scan obtained in step 3.1.2 The pose matrix Point cloud of standard physical tool in the corresponding scanner coordinate system , calculated robot The position transfer matrix of scanner C during scanning for: ； Further calculation shows that during the scanning process, the robot The standard physical tool point cloud below is: .

7. The method for automatic positioning of a robot tool coordinate system according to claim 6, wherein: The step 4 further comprises the following steps: Step 4.1: Obtain the point cloud data of the physical tool to be measured and the corresponding robot pose during the actual processing process, and convert the point cloud data of the physical tool to be measured from the scanner coordinate system to the robot coordinate system; Step 4.1.1: Robot moves to position , the scanner collects the point cloud data of the physical tool to be measured at this position; each point of the point cloud data is , then there is a corresponding point cloud ; Step 4.1.2: Repeat step 4.1.1 until you get Point cloud data The corresponding robot position ; Step 4.1.3: Summarize the robot's position at the point during the scan obtained in step 4.1.2 The pose matrix and the point cloud of the physical tool to be measured in the corresponding scanner coordinate system , the point cloud of the physical tool to be measured in the robot coordinate system during the scanning process is calculated as: ； Step 4.2: Obtain the tool coordinate system for visual calibration in the robot coordinate system , realize the positioning of the tool; Step 4.2.1: N The tool point cloud obtained in the robot coordinate system during the scanning process is filtered, cropped and merged to obtain the point cloud of the standard physical tool under the robot: ； in, Represents the original point cloud The result after filtering and cropping is that only valid matching point clouds are retained and noise and external interferences are deleted; Step 4.2.2: Obtain the standard physical tool model in the tool design model coordinate system , combined with the standard physical-design model conversion matrix The calculated point cloud of the standard physical tool in the robot coordinate system is: ； Step 4.2.3: Apply the ICP algorithm to calculate the transfer matrix from the point cloud of the standard physical tool model in the robot coordinate system to the point cloud of the physical tool to be measured: ； Step 4.2.4: Transfer matrix from the point cloud of the standard physical tool model in the robot coordinate system to the point cloud of the physical tool to be measured Standard physical object-design model conversion matrix , calculate the tool coordinate system matrix of the visual calibration in the robot coordinate system: .

Images