CN115056213A - Robot track self-adaptive correction method for large complex component - Google Patents
Robot track self-adaptive correction method for large complex component Download PDFInfo
- Publication number
- CN115056213A CN115056213A CN202210148528.8A CN202210148528A CN115056213A CN 115056213 A CN115056213 A CN 115056213A CN 202210148528 A CN202210148528 A CN 202210148528A CN 115056213 A CN115056213 A CN 115056213A
- Authority
- CN
- China
- Prior art keywords
- track
- point
- registration
- point cloud
- robot
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1656—Programme controls characterised by programming, planning systems for manipulators
- B25J9/1664—Programme controls characterised by programming, planning systems for manipulators characterised by motion, path, trajectory planning
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J11/00—Manipulators not otherwise provided for
- B25J11/0075—Manipulators for painting or coating
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Robotics (AREA)
- Mechanical Engineering (AREA)
- Numerical Control (AREA)
Abstract
The invention discloses a robot track self-adaptive correction method facing a large-scale complex component, which comprises deformation area identification and registration processing, track self-adaptive correction and posture self-adaptive correction. The invention carries out self-adaptive correction of the track and the attitude by combining an off-line programming method and a point cloud on-site scanning method. The problem that the off-line processing program of the large-scale component is difficult to match with the on-site component is solved, manual track correction is replaced, rigid registration and non-rigid registration are combined, and the calculation speed of on-site data and off-line data registration is improved. According to the method, a part with large local deformation is identified and registered by combining rigid registration and non-rigid registration, and model correction is rapidly carried out; the track self-adaptive function is realized by quickly matching the corresponding correction scheme through the process rule of the track knowledge base; and the attitude self-adaption function is realized by correcting the reachability check of the track points. And finally, outputting the whole set of track and pose.
Description
Technical Field
The invention relates to an off-line track self-adaptive correction method. In particular to a track posture self-adaptive correction method for a large-scale complex component.
Background
With the development of intelligent manufacturing technology, more and more factories requiring a spraying process start to use an automatic production line, and offline trajectory planning plays a great role, and the technology is continuously mature. Including off-line programming, trajectory modification, off-line simulation, etc. And the optimization and correction of the track are key technologies in the off-line programming system of the spraying robot.
In the field of complex large-scale component production, due to component deformation caused by a previous process, a large difference exists between component actual point cloud and a theoretical point cloud, and the registration between the standard point cloud and the measured point cloud from the point cloud of the digital analog is possibly difficult to complete only by rotating and translating the whole point cloud, namely the rigid registration method is possibly invalid. Therefore, a non-rigid registration method is introduced, even if each point has a scaling factor on three coordinates, local scaling is realized, and the effect of accurate registration is achieved. However, the non-rigid registration method mentioned above has a relatively large calculation amount, which results in a too long matching time and fails to achieve the real-time characteristic of the processing process. Secondly, in most cases, engineers use the existing experience, knowledge reserves and on-site conditions to modify the offline trajectory and attitude. Only the engineer corrects the attitude and the track according to the on-site observation of own experience knowledge, which causes the reduction of the process precision, and secondly, different engineers have different correction schemes and are difficult to maintain in the later period.
Disclosure of Invention
The invention aims to solve the technical problems in the prior art, and provides a robot track self-adaptive correction method for large complex components.
In order to achieve the purpose, the invention adopts the following technical scheme:
a track posture self-adaptive correction method for a large-scale complex component comprises the following operation steps:
1) generating an offline track:
establishing a workpiece coordinate system, a robot coordinate system and a tool coordinate system, importing a standard digital model of a product into a computer, and obtaining a point cloud file through discrete processing; generating a track and a posture as a standard registration pose of a subsequent step according to a process rule in a manual teaching or automatic off-line generation mode;
2) collecting and preprocessing component point cloud:
the method comprises the following steps that a multi-view laser sensor carries out three-dimensional point cloud collection on a component, a workpiece motion and camera static scheme is adopted, a ground rail chain plate conveying line is selected, and component scanning point cloud data are obtained through a point cloud splicing technology; meanwhile, dense point clouds are preprocessed, and evaluation indexes based on point cloud registration accuracy and registration speed are constructed to find a point cloud data geometric structure which can not be damaged, retain the local detail characteristics of components and effectively reduce the scale and density of the point clouds;
3) and (3) deformation region identification and registration processing:
firstly, carrying out noise reduction and downsampling on scanning, then carrying out coarse registration by using a rigid registration method, identifying a part with larger deformation, reducing the calculated amount during fine registration, saving registration time, particularly adopting Euclidean distance as a deformation evaluation standard, setting a threshold, dividing a point cloud into regions and different robot working ranges according to structural characteristics, dividing the point cloud into a plurality of different regions, calculating the relation between an average distance and the threshold to judge whether the region needs to be subjected to non-rigid registration, and if the average distance is larger than the threshold, carrying out integral non-rigid registration on the region; if the area is smaller than the threshold value, maintaining a rigid registration result in the area; then, fine registration is realized according to a non-rigid registration method, each registration point in a deformation area has a telescopic factor, finally, point clouds subjected to rigid registration and non-rigid registration are spliced to obtain transformation factors of all the point clouds, and the transformation factors are used as transformation matrixes in subsequent steps, so that track and posture self-adaptive correction is carried out;
4) track self-adaptive correction:
considering that the deformation of the large-scale component is the superposition state of shearing, torsion, bending and stretching deformation, the track self-adaptation is divided into two steps: firstly, neglecting telescopic deformation, only considering shearing, torsion and bending deformation in a deformation area, integrally mapping a standard track of an original point cloud to an actual point cloud by using a transformation function, and correcting and optimizing the track shape according to the requirements of a processing component; then, considering stretching deformation, the standard operation track of the robot may become too sparse or dense to meet the process requirements, the standard track needs to be densified, then an interpolation method is used for selecting appropriate track points again, the beat change is considered, the beat of the robot is recalculated according to the number of new tracks, and the running speed is adjusted;
5) posture self-adaptive correction:
firstly, giving an initial attitude to a track point, wherein the specific assignment scheme is as follows:
and taking half of the track point distance around the registration track point as a radius, taking the average normal vector of the partial point cloud as a reference datum, and then performing deflection correction on the basis according to the process parameters to serve as an initial attitude. Subsequently, the robot track corrected by the component deformation amount may have a condition that a part of postures are inaccessible due to the fact that inverse kinematics does not have a solution, and therefore, an iteration-based robot solving inverse solution method is adopted for verifying whether the newly generated track can meet the accessibility of the robot while meeting the process rationality. If not, an iteration mechanism is set to continuously optimize the robot track according to the process rule until the output meets the process requirement and the robot accessibility requirement is met to correct the robot operation track.
Preferably, the specific method in step 4) is as follows:
aiming at the robot track correction of shearing, twisting and bending deformation, after the registration transformation matrix is obtained in the step 3), the standard track points generated in the step 2) are transformed, and the position and the posture of the registered track points are calculated; at the moment, the new track is divided into characteristic areas according to the process knowledge base and is corrected, wherein the shape of the track point is adjusted, the distance between the original tracks changes in the registration process and becomes inconsistent with the spraying standard, the new track is matched with the corresponding rule again from the process knowledge base at the moment, the number of operation tracks is adjusted again, and the correction is completed. When the component is stretched and deformed, the point cloud density obtained by actual component scanning is obviously higher than that of the model offline point cloud, and the new track points generated by the original track points through the transformation function become sparse, so that the precision is reduced. The project adopts a point cloud slicing technology, slicing is carried out on actual scanning point cloud according to a mapped track, so that a continuous track point on the scanning point cloud surface is obtained, interpolation is continuously carried out to obtain dense points, and track correction is carried out according to a robot track correction method for shearing, torsion and bending deformation.
Preferably, the specific attitude optimization calculation method in step 5) is as follows:
and evaluating the attitude of the unreachable point combined with the previous track point and the next track, judging whether the attitude is the initial point of the spraying transition point, and if so, judging whether the attitude is the initial point of the spraying transition point: copying the previous gesture; if not: taking the median vector (Z) of the spraying direction between the upper and lower points i-1 +Z i+1 ) Taking the intermediate vector as the spraying direction of the point, continuously finding the gesture with inverse solution clockwise and anticlockwise in the same step length around the intermediate vector, and if the gesture exists at the same time, calculating the sum (theta) of the joint angles of the robot in two conditions 1 +θ 2 +…+θ n ) And taking the smaller value as the optimal solution.
Compared with the prior art, the invention has the following obvious prominent substantive characteristics and remarkable advantages:
1. the method of the invention uses rigid registration to roughly position the deformation area, and non-rigid registration to precisely match the scanning point cloud and the standard point cloud, thereby improving the sensing precision of parts, not only ensuring the deformation characteristics to be accurately positioned, but also reducing the calculation amount and shortening the calculation time; the track self-adaptive technology converts the experience of an engineer into information which can be quantitatively described, and a track adaptive scheme is normalized; the gesture self-adaption technology verifies whether the final track points have inverse solutions or not, and ensures that all track points can reach, so that a complete correction track is generated;
2. the invention carries out self-adaptive correction of the track and the attitude by combining an off-line programming method and a point cloud on-site scanning method; the problem that the off-line processing program of the large-scale component is difficult to match with the on-site component is solved, manual track repair is replaced, rigid registration and non-rigid registration are combined, and the calculation speed of on-site data and off-line data registration is improved;
3. according to the invention, on-site part scanning perception is adopted to replace the assessment of engineers, so that the feature extraction is more accurate; by combining rigid registration with non-rigid registration, on the basis of keeping most of the original tracks, identifying and registering the part with large local deformation, and quickly correcting the model; the track self-adaptive function is realized by quickly matching the corresponding correction scheme through the process rule of the track knowledge base; and searching an optimal solution in all reachable solutions aiming at the unreachable points by correcting the reachability check of the track points, thereby realizing the posture self-adaption function. And finally, outputting the whole set of track and pose.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a flowchart of a deformed region identification and registration process according to a preferred embodiment of the present invention.
FIG. 3 is a flow chart of attitude adaptive correction in accordance with a preferred embodiment of the present invention.
Detailed Description
The invention is described in detail with respect to preferred embodiments in conjunction with the accompanying drawings:
the first embodiment is as follows:
referring to fig. 1, a robot track adaptive correction method for a large-scale complex component includes the following operation steps:
1) generating an offline track:
establishing a workpiece coordinate system, a robot coordinate system and a tool coordinate system, importing a standard digital analogy of a product into a computer, and obtaining a point cloud file through discrete processing; generating a track and a posture as a standard registration pose of a subsequent step according to a process rule in a manual teaching or automatic off-line generation mode;
2) collecting and preprocessing component point cloud:
the method comprises the following steps of utilizing a multi-view laser sensor to conduct three-dimensional point cloud collection on a component, adopting a scheme of workpiece movement and camera stillness, selecting a ground rail chain plate conveying line, and obtaining component scanning point cloud data through a point cloud splicing method; meanwhile, dense point clouds are preprocessed, and evaluation indexes based on point cloud registration accuracy and registration speed are constructed to find a point cloud data geometric structure which can not be damaged, retain the local detail characteristics of components and effectively reduce the scale and density of the point clouds;
3) and (3) deformation region identification and registration processing:
the computer inputs a digital-analog, an off-line track and a field scanning point cloud, and preprocesses the field scanning point cloud to reduce the sample size and remove noise, and then performs coarse registration by using a rigid registration method to identify a part with larger deformation, reduce the calculated amount during fine registration and save the registration time; specifically, the Euclidean distance is used as a deformation evaluation standard, a threshold value is set, the point cloud is divided into a plurality of different areas according to the structural characteristics of the area and the working ranges of different robots, the relation between the average distance and the threshold value is calculated to judge whether the area needs to be subjected to non-rigid registration, and if the average distance is larger than the threshold value, the area is subjected to integral non-rigid registration; if the area is smaller than the threshold value, maintaining a rigid registration result in the area;
then, fine registration is realized according to a non-rigid registration method, each registration point in a deformation area has a telescopic factor, finally, point clouds subjected to rigid registration and non-rigid registration are spliced to obtain transformation factors of all the point clouds, and the transformation factors are used as transformation matrixes in subsequent steps, so that track and posture self-adaptive correction is carried out;
4) track self-adaptive correction:
considering that the deformation of the large-scale component is the superposition state of shearing, torsion, bending and stretching deformation, the track self-adaptation is divided into two steps:
firstly, neglecting telescopic deformation, only considering shearing, torsion and bending deformation in a deformation area, integrally mapping a standard track of an original point cloud to an actual point cloud by using a transformation function, and correcting and optimizing the track shape according to the requirements of a processing component;
then, considering stretching deformation, the standard operation track of the robot may become too sparse or dense and not meet the process requirements, the standard track needs to be densified first, then an interpolation method is used for re-selecting proper track points, the change of the beat is considered, the beat of the robot is recalculated according to the number of new tracks, and the running speed is adjusted;
5) posture self-adaptive correction:
firstly, giving an initial attitude to a track point, wherein the specific assignment scheme is as follows:
taking half of the distance between the track points as a radius around the registration track points, taking an average normal vector of the partial point cloud as a reference standard, and then performing deflection correction on the basis according to process parameters to serve as an initial attitude;
then, considering the condition that the robot track corrected by the component deformation amount possibly has part of postures and the robot is unreachable due to the fact that inverse kinematics does not have a solution, an iterative-based robot solving inverse solution method is adopted to verify whether the newly generated track meets the process rationality and can meet the robot reachability; if not, an iteration mechanism is set to continuously optimize the robot track according to the process rule until the output meets the process requirement and the robot accessibility requirement is met to correct the robot operation track.
In the method, the rigid registration coarse positioning deformation area is used, the non-rigid registration fine matching scanning point cloud and the standard point cloud is adopted, the part sensing precision is improved, the deformation characteristics can be accurately positioned, the calculated amount can be reduced, and the calculation time can be shortened; the track self-adaptive method is adopted to convert the experience of an engineer into information which can be quantitatively described, and a track adaptive scheme is normalized; the gesture self-adaption method verifies whether the final track points have inverse solutions or not, and ensures that all track points can reach, so that a complete correction track is generated.
Example two:
this embodiment is substantially the same as the first embodiment, and is characterized in that:
in this embodiment, in the step 4), after the registration transformation matrix is obtained in the step 3) for shearing, twisting, and bending deformation in the comprehensive deformation component, the standard trajectory point is transformed, and the position and the posture of the registered trajectory point are calculated; at the moment, dividing a new track into characteristic areas according to a process knowledge base and correcting the characteristic areas, wherein the shape of each track is adjusted, and after non-rigid registration, the original linear track may generate a jittering track after transformation, so that the shape of the track needs to be adjusted according to the conditions to meet the working requirements; the large-scale component is difficult to deform during processing so as to generate manufacturing errors, the actual component scans the condition that the point cloud density is obviously more than or sparse than the model offline point cloud, and a new track point generated by the original track point through a transformation function becomes sparse or dense, so that the precision is reduced or the processing beat is lengthened; the method comprises the steps of adopting a point cloud slicing technology, selecting a proper reference point on an actual scanning point cloud according to a mapped track for slicing to obtain a continuous track on the surface of the scanning point cloud, then continuously adjusting the angle of a tangent plane in the gradient reducing direction by using the distance sum of the track point and the tangent plane as an evaluation standard to ensure that the precision reaches a required range, selecting the tangent plane at the moment to process a model to obtain an optimal track, continuously carrying out interpolation on the track to obtain posture and position dense points, and determining the density of track points according to requirements and selecting the track points.
In this embodiment, the specific attitude optimization calculation method in step 5) includes:
and evaluating the attitude of the unreachable point combined with the previous track point and the next track, judging whether the attitude is the initial point of the spraying transition point, and if so, judging whether the attitude is the initial point of the spraying transition point: copying the Y axis of the advancing axis of the tool coordinate system of the previous point posture, fixing the Z axis direction, updating the X axis, recalculating the Y axis, and determining the posture of the point; if not: taking the median vector (Z) of the spraying direction between the upper and lower points i-1 +Z i+1 ) Taking the intermediate vector as the spraying direction of the point, continuously finding the gesture with inverse solution clockwise and anticlockwise in the same step length around the intermediate vector, and if the gesture exists at the same time, calculating the sum (theta) of the joint angles of the robot in two conditions 1 +θ 2 +…+θ n ) Taking the smaller value as the optimal solution, (theta) 1 +θ 2 +…+θ n ) Representing each joint angle.
The embodiment carries out self-adaptive correction on the track and the posture by combining an off-line programming method and a point cloud on-site scanning method; the problem that the off-line processing program of the large-scale component is difficult to match with the on-site component is solved, manual track repair is replaced, rigid registration and non-rigid registration are combined, and the calculation speed of on-site data and off-line data registration is improved; the invention carries out self-adaptive correction of the track and the attitude by combining an off-line programming method and a point cloud on-site scanning method; the problem that the off-line processing program of the large-scale component is difficult to match with the on-site component is solved, manual track repair is replaced, rigid registration and non-rigid registration are combined, and the calculation speed of on-site data and off-line data registration is improved; in the embodiment, by combining rigid registration with non-rigid registration, on the basis of keeping most of the original tracks, a part with large local deformation is identified and registered, and model correction is performed quickly; the track self-adaptive function is realized by quickly matching the corresponding correction scheme through the process rule of the track knowledge base; and searching an optimal solution in all reachable solutions aiming at the unreachable points by correcting the reachability check of the track points, thereby realizing the posture self-adaption function. And finally, outputting the whole set of track and pose.
Example three:
referring to fig. 1, a robot track adaptive correction method for a large-scale complex component includes the following steps:
1) generating an offline track: and establishing a DH model according to a robot, establishing a workpiece coordinate system, a robot coordinate system and a tool coordinate system, naming the product digifax, introducing the product digifax into a computer, and obtaining a point cloud file through discrete processing. And generating the track and the posture according to the process rule in a manual teaching or automatic off-line generation mode.
2) Collecting and preprocessing component point cloud: the multi-view laser sensor carries out three-dimensional point cloud collection on the component, a workpiece motion and camera static scheme is adopted, a ground rail chain plate conveying line is selected, and component scanning point cloud data are obtained through a point cloud splicing technology. Meanwhile, dense point cloud is preprocessed, and evaluation indexes based on point cloud registration accuracy and registration speed are constructed to find a point cloud data geometric structure which can not be damaged, retain local detail features of components and effectively reduce the point cloud scale and density.
3) And (3) deformation region identification and registration processing: firstly, a neighbor searching method of a KD tree based on a point cloud data topological structure utilizes a radius filter to remove outliers efficiently, then a rigid registration method is used for coarse registration, a part with a large deformation is identified, the calculation amount during fine registration is reduced, the registration time is saved, specifically, an Euclidean distance is used as a deformation evaluation standard, a threshold value is set, the point cloud is subjected to region division according to structural characteristics, and whether non-rigid registration is required in the region is judged according to the relation between the average distance and the threshold value. And then, realizing fine registration according to a non-rigid registration method, wherein each registration point in the deformation region has a scaling factor, and the scaling factor is used as a transformation matrix in the subsequent steps so as to perform self-adaptive correction on the track and the posture.
4) Track self-adaptive correction: and aiming at shearing, twisting and bending deformation in the deformation area, the change of the density of the point cloud can be not considered, the original point cloud and the track are integrally mapped onto the actual point cloud by using a transformation function, and the track and the posture are corrected and optimized according to a process rule. However, when the component is stretched and deformed, the standard operation track of the robot may become too sparse to meet the process requirements, and the standard track needs to be subjected to densification first, and then the posture of the robot needs to be optimized by using an interpolation method.
5) Posture self-adaptive correction: the robot trajectory corrected according to the amount of deformation of the member may have a case where the robot is unreachable due to partial attitude being unsolved by inverse kinematics. For this purpose, an iterative-based inverse solution method for robot determination is used to check whether the newly generated trajectory is able to satisfy the accessibility of the robot while satisfying the process rationality. If not, an iteration mechanism is set to continuously optimize the robot track according to the process rule until the output meets the process requirement and the robot accessibility requirement is met to correct the robot operation track.
The specific method in the step 4) is as follows: the robot used in the examples was IRB5400, the spray tool used a non-spherical wrist, and its DH parameters are as follows:
TABLE 1 DH parameters table
| i | θ i | d i | a i-1 | α i-1 |
| 1 | θ 1 | d 1 | 0 | 0 |
| 2 | θ 2 | 0 | a 1 | -90° |
| 3 | θ 3 | 0 | a 2 | 0 |
| 4 | θ 4 | d 4 | a 3 | -90° |
| 5 | θ 5 | d 5 | 0 | β |
| 6 | -θ 5 | d 6 | 0 | -2β |
| 7 | θ 6 | d 7 | 0 | β |
Aiming at shearing, torsion and bending deformation in the comprehensive deformation component, after the registration transformation matrix is obtained in the step 3), the standard track points generated in the step 2) are transformed, and the position and the posture of the registered track points are calculated. At the moment, the new trajectory is divided into characteristic regions according to a process knowledge base and is corrected, wherein the shape of each trajectory is adjusted, after non-rigid registration, the original linear trajectory may generate a jittering trajectory after transformation, and therefore, the shape of the trajectory needs to be adjusted according to the above conditions, so that the working requirements are met. The large-scale component is difficult to deform during processing so as to generate manufacturing errors, the actual component scans the condition that the point cloud density is obviously more than or sparsely than the model offline point cloud, and the new track points generated by the original track points through the transformation function become sparse or dense, so that the precision is reduced or the processing beat is lengthened. The invention adopts a point cloud slicing technology, selects a proper reference point on the actual scanning point cloud according to the mapped track for slicing to obtain a continuous track on the surface of the scanning point cloud, then uses the distance sum of the track point and a tangent plane as an evaluation standard, continuously adjusts the tangent plane angle in the gradient reducing direction to ensure that the precision reaches the required range, selects the tangent plane at the moment to process the model to obtain an optimal track, continuously performs interpolation on the track to obtain posture and position dense points, and determines the track point density according to the requirement and selects the track point density.
The specific attitude optimization calculation method in the step 5) comprises the following steps: firstly, outputting a modified track, taking a row of track points as an example for analysis, wherein the specific positions and postures are as follows in the following table 2:
TABLE 2 detailed position and attitude information Table
Calculating unreachable points, evaluating the gesture of the track point 104 combining the previous track point and the next track, and judging whether the gesture is the initial point of the spraying transition point, wherein the gesture is not the transition point: taking the median vector (Z) of the spraying direction between the upper and lower points i-1 +Z i+1 ) Namely (1.73205-6.00926e-171), taking the intermediate vector as the spraying direction of the point, continuously searching the gesture with the inverse solution by taking 1 degree as a step length clockwise and anticlockwise around the point, finding the inverse solution when rotating by 7 degrees anticlockwise and not having the inverse solution by 7 degrees clockwise, taking the gesture at the moment as the optimal gesture and updating the track.
The embodiment of the invention is a robot track self-adaptive correction method for large complex components, which comprises the steps of firstly identifying and registering a deformation area, then carrying out track self-adaptive correction and then carrying out attitude self-adaptive correction. The embodiment of the invention carries out the self-adaptive correction of the track and the posture by combining the off-line programming and the point cloud on-site scanning method. The embodiment solves the problem that the large-scale component offline processing program is difficult to match with the field component, replaces manual track correction, and improves the calculation speed of field data and offline data registration by combining rigid registration and non-rigid registration. According to the embodiment of the invention, the on-site part scanning perception is adopted to replace the evaluation of engineers, so that the feature extraction is more accurate; by combining rigid registration with non-rigid registration, on the basis of keeping most of the original tracks, identifying and registering the part with large local deformation, and quickly correcting the model; in the embodiment, the track self-adaptive function is realized by quickly matching the corresponding correction scheme through the process rule of the track knowledge base; and searching an optimal solution in all reachable solutions aiming at the unreachable points by correcting the reachability check of the track points, thereby realizing the posture self-adaption function. And finally, outputting the whole set of track and pose.
The embodiments of the present invention have been described with reference to the accompanying drawings, but the present invention is not limited to the embodiments, and various changes and modifications can be made according to the purpose of the invention, and any changes, modifications, substitutions, combinations or simplifications made according to the spirit and principle of the technical solution of the present invention shall be equivalent substitutions, as long as the purpose of the present invention is met, and the present invention shall fall within the protection scope of the present invention without departing from the technical principle and inventive concept of the present invention.
Claims (3)
1. A robot track self-adaptive correction method for large-scale complex components is characterized by comprising the following operation steps:
1) generating an offline track:
establishing a workpiece coordinate system, a robot coordinate system and a tool coordinate system, importing a standard digital analogy of a product into a computer, and obtaining a point cloud file through discrete processing; generating a track and a posture as a standard registration pose of a subsequent step according to a process rule in a manual teaching or automatic off-line generation mode;
2) collecting and preprocessing component point cloud:
the method comprises the following steps of utilizing a multi-view laser sensor to conduct three-dimensional point cloud collection on a component, adopting a scheme of workpiece movement and camera stillness, selecting a ground rail chain plate conveying line, and obtaining component scanning point cloud data through a point cloud splicing method; meanwhile, dense point clouds are preprocessed, and evaluation indexes based on point cloud registration accuracy and registration speed are constructed to find a point cloud data geometric structure which can not be damaged, retain the local detail characteristics of components and effectively reduce the scale and density of the point clouds;
3) and (3) deformation region identification and registration processing:
the computer inputs a digital-analog, an off-line track and a field scanning point cloud, the field scanning point cloud is preprocessed, so that the sample size is reduced, noise is removed, then a rigid registration method is used for rough registration, a part with larger deformation is identified, the calculation amount during fine registration is reduced, and registration time is saved; specifically, the Euclidean distance is used as a deformation evaluation standard, a threshold value is set, the point cloud is divided into a plurality of different areas according to the structural characteristics of the area and the working ranges of different robots, the relation between the average distance and the threshold value is calculated to judge whether the area needs to be subjected to non-rigid registration, and if the average distance is larger than the threshold value, the area is subjected to integral non-rigid registration; if the area is smaller than the threshold value, maintaining a rigid registration result in the area;
then, fine registration is realized according to a non-rigid registration method, each registration point in a deformation area has a telescopic factor, finally point clouds of rigid registration and non-rigid registration are spliced to obtain transformation factors of all the point clouds, and the point clouds are used as transformation matrixes in the subsequent steps, so that track and posture self-adaptive correction is carried out;
4) track self-adaptive correction:
considering that the deformation of the large-scale component is the superposition state of shearing, torsion, bending and stretching deformation, the track self-adaptation is divided into two steps:
firstly, ignoring stretching deformation, only considering shearing, twisting and bending deformation in a deformation area, integrally mapping a standard track of an original point cloud onto an actual point cloud by using a transformation function, and correcting and optimizing the track shape according to the requirements of a processing component;
then, considering stretching deformation, the standard operation track of the robot may become too sparse or dense and not meet the process requirements, the standard track needs to be densified first, then an interpolation method is used for re-selecting proper track points, the change of the beat is considered, the beat of the robot is recalculated according to the number of new tracks, and the running speed is adjusted;
5) posture self-adaptive correction:
firstly, giving an initial attitude to a track point, wherein the specific assignment scheme is as follows:
taking half of the track point distance around the registration track point as a radius, taking the average normal vector of the partial point cloud as a reference datum, and then carrying out deflection correction on the basis according to process parameters to be used as an initial attitude;
then, considering the condition that the robot track corrected by the component deformation amount possibly has part of postures and is unreachable due to the fact that inverse kinematics does not have a solution, an iterative-based robot solving inverse solution method is adopted and used for verifying whether the newly generated track meets the process rationality and can also meet the robot reachability; if not, an iteration mechanism is set to continuously optimize the robot track according to the process rule until the output meets the process requirement and the robot accessibility requirement is met to correct the robot operation track.
2. The large-scale complex component-oriented robot track self-adaptive correction method according to claim 1, characterized in that: in the step 4), aiming at shearing, twisting and bending deformation in the comprehensive deformation component, after the registration transformation matrix is obtained in the step 3), the standard track point is transformed, and the position and the posture of the registered track point are calculated; at the moment, dividing a new track into characteristic areas according to a process knowledge base and correcting the characteristic areas, wherein the shape of each track is adjusted, and after non-rigid registration, the original linear track may generate a jittering track after transformation, so that the shape of the track needs to be adjusted according to the conditions to meet the working requirements; the large-scale component is difficult to deform during processing so as to generate manufacturing errors, the actual component scans the condition that the point cloud density is obviously more than or sparse than the model offline point cloud, and a new track point generated by the original track point through a transformation function becomes sparse or dense, so that the precision is reduced or the processing beat is lengthened; a point cloud slicing method is adopted, a proper reference point is selected on an actual scanning point cloud according to a mapped track for slicing to obtain a continuous track on the surface of the scanning point cloud, then the angle of a tangent plane is continuously adjusted in the gradient reducing direction by using the distance sum of the track point and the tangent plane as an evaluation standard to ensure that the precision reaches a required range, the tangent plane at the moment is selected to process a model to obtain an optimal track, interpolation is continuously carried out on the track to obtain posture and position dense points, and the density of track points is determined and selected according to requirements.
3. The large-scale complex component-oriented robot track self-adaptive correction method according to claim 1, characterized in that: in the step 5), a specific attitude optimization calculation method is as follows:
evaluating the attitude of the unreachable point combined with the previous track point and the next track to judge whether the unreachable point is the starting point of the spraying transition point; if yes, copying a Y axis of an advancing axis of a tool coordinate system of the previous point posture, fixing the direction of a Z axis, updating an X axis, recalculating the Y axis, and determining the posture of the point; if not, the intermediate vector (Z) of the spraying direction is taken between the upper point and the lower point i-1 +Z i+1 ) Taking the intermediate vector as the spraying direction of the point, continuously finding the gesture with inverse solution clockwise and anticlockwise in the same step length around the intermediate vector, and if the gesture exists at the same time, calculating the sum (theta) of the joint angles of the robot in two conditions 1 +θ 2 +…+θ n ) Taking the smaller value as the optimal solution, (theta) 1 +θ 2 +…+θ n ) Representing each joint angle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210148528.8A CN115056213A (en) | 2022-02-18 | 2022-02-18 | Robot track self-adaptive correction method for large complex component |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210148528.8A CN115056213A (en) | 2022-02-18 | 2022-02-18 | Robot track self-adaptive correction method for large complex component |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115056213A true CN115056213A (en) | 2022-09-16 |
Family
ID=83197192
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210148528.8A Pending CN115056213A (en) | 2022-02-18 | 2022-02-18 | Robot track self-adaptive correction method for large complex component |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115056213A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116476070A (en) * | 2023-05-22 | 2023-07-25 | 北京航空航天大学 | Method for adjusting scanning measurement path of large-scale barrel part local characteristic robot |
-
2022
- 2022-02-18 CN CN202210148528.8A patent/CN115056213A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116476070A (en) * | 2023-05-22 | 2023-07-25 | 北京航空航天大学 | Method for adjusting scanning measurement path of large-scale barrel part local characteristic robot |
| CN116476070B (en) * | 2023-05-22 | 2023-11-10 | 北京航空航天大学 | Method for adjusting scanning measurement path of large-scale barrel part local characteristic robot |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110227876B (en) | Robot welding path autonomous planning method based on 3D point cloud data | |
| CN109976259B (en) | VTK-based robot free-form surface workpiece polishing offline programming method | |
| CN112348864B (en) | Three-dimensional point cloud automatic registration method for laser contour features of fusion line | |
| CN110069041B (en) | Workpiece machining method and system based on-machine measurement | |
| US20230400830A1 (en) | Methods and systems of fast optimization and compensation for volumetric positioning errors of rotary axes of five-axis cnc machine tools | |
| CN109683552B (en) | Numerical control machining path generation method on complex point cloud model guided by base curve | |
| CN102785166B (en) | Kinematic transformation based grinding machining method for numerically controlled grinding wheel | |
| CN113741426B (en) | Robot machining path planning method based on local point cloud curve fitting | |
| CN112223294A (en) | Mechanical arm machining track correction method based on three-dimensional vision | |
| CN111975767B (en) | Multi-robot visual detection system collaborative motion planning method based on multi-stage task allocation | |
| CN110837715B (en) | Complex curved surface machining error compensation method based on reverse engineering technology | |
| CN114571153A (en) | Weld joint identification and robot weld joint tracking method based on 3D point cloud | |
| CN111820545A (en) | Method for automatically generating sole glue spraying track by combining offline and online scanning | |
| CN112947298A (en) | Method, system and terminal for optimally generating curved surface machining track of robot | |
| CN114018155A (en) | Method and system for detecting precision of chemical milling laser engraving profile | |
| CN113587807A (en) | Scanning path generation method for aircraft wall panel machining feature surface | |
| CN115056213A (en) | Robot track self-adaptive correction method for large complex component | |
| CN111958602A (en) | Real-time inverse solution method for wrist offset type 6-axis robot | |
| CN109636077B (en) | Variable node assembly path planning method based on dual local pose transformation | |
| CN112906215B (en) | Pipe tool path generation method based on SolidWorks secondary development | |
| CN109397293B (en) | Ground level error modeling and compensating method based on mobile robot | |
| CN112991187B (en) | Convolution twin-point network blade profile splicing system based on multiple spatial similarities | |
| CN114036634A (en) | Thin-wall part self-adaptive trimming method based on flexible clamping | |
| CN113111553A (en) | Interpolation deformation grid-based large deformation motion numerical simulation method | |
| CN110458822B (en) | Non-contact three-dimensional matching detection method for complex curved surface part |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |