A visual guidance-based automatic calibration method for a scribing robot

Abstract

The present application relates to the technical field of image analysis, in particular to a kind of visual guidance-based scribing robot automatic calibration method, comprising the following steps: based on the image data of work site obtained by scribing robot, establish the image set including different resolution levels, carry out corner point detection and straight line segment detection on each level image in parallel.The present application extracts the corner point and straight line segment features under different scales simultaneously through the parallel detection mechanism of multi-resolution level image set, constructs the multi-scale feature point set of high robustness by combining cross-level coordinate stability measurement and response intensity quantification.Based on preset geometric constraint condition, feature points are dynamically screened and grouped, noise interference and false detection features are removed, and candidate calibration structure set with spatial consistency is generated.Weighted contribution value fitting is used to comprehensively consider the cross-scale stability and detection confidence of feature points, so as to improve the geometric precision and anti-interference ability of calibration reference point coordinates.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image analysis, in particular to a visual guidance-based automatic calibration method for a marking robot. BACKGROUND

[0002] The technical field of image analysis focuses on extracting, analyzing and quantifying information in digital images through algorithms and computational models, covering core directions such as image enhancement, feature detection, target recognition, and spatial measurement.

[0003] Existing image analysis techniques are difficult to effectively deal with the problem of feature missing or false detection caused by scale changes in complex scenes, for example, low-resolution images may lose small corner points, and high-resolution images are easily disturbed by noise. Traditional geometric constraints are mostly static threshold settings, lacking dynamic adjustment ability for marking tasks, which may cause effective features to be mistakenly deleted or redundant points to be retained during feature screening. Therefore, improvement is needed. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and to propose a visual guidance-based automatic calibration method for a marking robot.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme, a visual guidance-based automatic calibration method for a marking robot, comprising the following steps:

[0006] Based on the work site image data obtained by the marking robot, an image set including different resolution levels is established, and corner point detection and straight line segment detection are performed on each level image in parallel to generate a multi-scale basic geometric feature point set;

[0007] Based on the multi-scale basic geometric feature point set, a preset geometric constraint condition for the marking task is called to determine the compliance of each feature point in the multi-scale basic geometric feature point set, to obtain a feature point subset that meets the constraint, and based on the feature point subset that meets the constraint, grouping and association are performed to establish a candidate calibration marker structure set;

[0008] Based on the candidate calibration marker structure set, the feature points in each structure are traversed, the detection response strength and the coordinate stability measurement value across the scale levels in the multi-scale basic geometric feature point set are quantified, the contribution weight parameter of each feature point is determined, the weighted feature point contribution value is obtained, and based on the weighted feature point contribution value, the feature point coordinates associated with each candidate structure are fitted to obtain the calibration reference point coordinates;

[0009] Based on the calibration reference point coordinates as a target position, instructing the scribe robot to perform a test scribe, synchronously collecting a splashing and scattering pattern image of the scribe robot nozzle end at the moment of contact with the working ground, obtaining the micro-mark pattern geometric features, based on the micro-mark pattern geometric features, adjusting the current position parameters of the tool end point of the scribe robot.

[0010] Preferably, the step of obtaining the multi-scale basic geometric feature point set is:

[0011] The working site image data is subjected to multi-level resolution reconstruction, and a Gaussian pyramid generation method is used to adjust the resolution layer by layer and superimpose it to the set to form a multi-resolution image set;

[0012] Based on the multi-resolution image set, corner point detection and straight line segment detection are performed on each level image in parallel, wherein the corner point detection calculates the corner point response value of each pixel based on the gradient intensity matrix, the straight line segment detection extracts the line segment end point through the edge direction histogram statistics and the line segment connectivity criterion, and the response value of the corner point and the line segment end point exceeding the dynamic threshold are screened as candidate geometric response points to generate a multi-level geometric response point set.

[0013] The coordinate interpolation and scale attribute calculation are performed on each candidate geometric response point in the multi-level geometric response point set, wherein the coordinate interpolation uses a quadratic polynomial fitting method to optimize the coordinate accuracy, and the scale attribute is determined based on the ratio of the level resolution and the original resolution of the geometric response point, to obtain a multi-scale basic geometric feature point set.

[0014] Preferably, the step of obtaining the feature point subset satisfying the constraint is:

[0015] Based on the preset scribe task geometric constraint condition parameter configuration table, the minimum angle and the maximum angle of the target angle range, the lower threshold of the number of collinear points, the minimum distance and the maximum distance of the point pair spacing are called;

[0016] Each feature point in the multi-scale basic geometric feature point set is traversed, the angle formed by the feature point and the adjacent two points is calculated, the number of collinear points continuously distributed on the straight line segment where the feature point is located is counted, and the Euclidean distance between the feature point and the nearest neighbor point is measured, and whether the angle formed by the feature point and the adjacent two points is within the minimum angle and the maximum angle interval, whether the number of collinear points is greater than the lower threshold of the number of collinear points, and whether the Euclidean distance is within the minimum distance and the maximum distance interval of the point pair spacing are judged, respectively, if any one of the judgment results is true, the feature point is marked as an effective candidate point, and a constraint compliance judgment result set is generated.

[0017] Based on the constraint compliance judgment result set, all feature points marked as effective candidate points are screened, and the feature points not marked are removed to form a feature point subset satisfying the constraint.

[0018] Preferably, the step of obtaining the candidate calibration marker structure set is:

[0019] Based on the preset corner point combination rule parameter table, the minimum allowed distance value between corner points, the maximum allowed distance value between corner points, the minimum allowed included angle value between corner points, the maximum allowed included angle value between corner points, the minimum allowed distance value between collinear points, the minimum required number of collinear points, and the fixed distance pairing allowed error value are called to construct a geometric grouping rule parameter set including corner point combination distance interval, corner point combination included angle range, collinear arrangement distance threshold and number threshold, and fixed distance pairing error threshold;

[0020] Traverse all feature points in the subset of feature points satisfying the constraints, and perform the following operations for each feature point: for the corner point combination rule, calculate the straight line distance value between the feature point and other feature points, judge whether the distance value is within the corner point combination distance interval, at the same time calculate the included angle value of the connecting line of the two points and the reference coordinate axis, judge whether the angle value is within the preset included angle range, and select the feature point pair that satisfies the distance interval and angle range conditions as the candidate corner point combination; for the collinear arrangement rule, search the distance value of adjacent feature points on the straight line segment where the feature point is located, count the number of feature points whose consecutive distances do not exceed the collinear arrangement distance threshold, and select the feature point sequence whose number reaches the minimum required number of collinear points as the candidate collinear group; for the fixed distance pairing rule, measure the actual distance value between the feature point and all feature points, and select the feature point pair whose absolute value of the difference between the actual distance and the preset fixed distance does not exceed the allowed error value as the candidate fixed distance pairing group, to generate a candidate grouping set containing the candidate corner point combination, the candidate collinear group and the candidate fixed distance pairing group;

[0021] Based on the candidate grouping set, in the case that the same feature point is assigned to multiple candidate groups, the group that meets the most grouping rules is preferentially retained, the groups that have coordinate overlap or logical conflict with other groups are eliminated, the remaining independent and conflict-free candidate corner point combination, candidate collinear group and candidate fixed distance pairing group are combined into a complete feature point combination, and a candidate calibration marker structure set is established.

[0022] Preferably, the step of obtaining the weighted feature point contribution value is:

[0023] Traverse each feature point in the candidate calibration marker structure set, extract the response intensity of the feature point in the corner point detection of each level image, and the coordinate data set at different resolution levels in the multi-scale basic geometric feature point set, to generate a feature point response intensity-multi-scale coordinate data set;

[0024] Based on the feature point response intensity-multi-scale coordinate data set, calculate the coordinate stability measure value of each feature point across the scale level to obtain a feature point coordinate stability set;

[0025] Based on the feature point response intensity-multiscale coordinate dataset and the feature point coordinate stability set, the weighted feature point contribution value is calculated.

[0026] Preferably, the step of obtaining the coordinates of the calibration reference point is as follows:

[0027] Traverse each candidate structure in the candidate calibration label structure set, extract the weighted feature point contribution value set and the corresponding feature point coordinate set of all feature points in the structure, and generate the candidate structure feature point contribution-coordinate dataset;

[0028] Based on the candidate structure feature point contribution-coordinate dataset, the calibration reference point fitting coordinates are calculated, and the calibration reference point coordinates of all candidate structures are generated.

[0029] Preferably, the step of obtaining the morphological geometric features of the micro-marker is as follows:

[0030] The coordinates of the calibration reference point are transmitted to the control console of the line drawing robot, triggering the robot to perform trial line drawing operations. Simultaneously, images of the splash dispersion pattern at the moment when the nozzle tip contacts the ground are acquired, generating a splash image sequence.

[0031] Based on the splash image sequence, the following processing is performed on each frame: the grayscale threshold is determined by the maximum inter-class variance method to complete the binarization segmentation, morphological closing operation is performed on the binary image to eliminate holes, the Canny operator is applied to extract the edge of the splash area, and a splash contour dataset is generated.

[0032] Iterate through the splatter profile dataset and perform geometric moment calculations for each closed profile point set: calculate the first moment M of the closed profile point set in the x-axis direction. 10 and the first moment M in the y-axis direction 01 With the zero-order moment M of the contour region 00 Based on x c =M 10 / M 00 With y c =M 01 / M 00 Obtain the physical coordinates x of the centroid c and y c Construct the geometric features of microscopic marker morphology.

[0033] Preferably, based on the geometric features of the micro-marker morphology, the step of calculating the spatial deviation vector between the centroid position coordinates and the calibration reference point coordinates, and adjusting the current position parameters and attitude parameters of the tool end point of the marking robot according to the spatial deviation vector is as follows:

[0034] Calculate the difference between the centroid physical coordinates and the preset coordinate components axis by axis to generate a deviation vector dataset including an x-axis deviation amount and a y-axis deviation amount based on the centroid physical coordinates of the micro-mark morphology geometric features;

[0035] According to the deviation vector dataset, decompose the deviation into a horizontal translation compensation step and a longitudinal translation compensation step in the robot base coordinate system to generate a set of pose correction parameters;

[0036] Based on the set of pose correction parameters, adjust the current position parameters of the tool end point of the line marking robot.

[0037] Compared with the prior art, the advantages and positive effects of the present application are:

[0038] The present application synchronously extracts corner point and straight line segment features at different scales through a parallel detection mechanism of a multi-resolution hierarchical image set, and constructs a high-robustness multi-scale feature point set by combining cross-hierarchical coordinate stability measurement and response intensity quantification. Based on preset geometric constraint conditions, the feature points are dynamically selected and grouped, noise interference and false detection features are removed, and a candidate calibration structure set with spatial consistency is generated. Weighted contribution value fitting is used to comprehensively consider the cross-scale stability and detection confidence of the feature points, thereby improving the geometric precision and anti-interference ability of the calibration reference point coordinates. Through the feedback of the micro-geometric features of the trial line marking splashing shape, the robot end position and attitude parameters are corrected in real time, forming a closed-loop calibration process. The calibration process has dynamic environmental adaptability, reduces the dependence on manual intervention, and improves the calibration precision. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 The present application is a step schematic diagram. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0041] Please refer to Figure 1 The present application provides a technical scheme, a line marking robot automatic calibration method based on visual guidance, comprising the following steps:

[0042] Based on the work site image data obtained by the line marking robot, an image set including different resolution levels is established, corner point detection and straight line segment detection are performed on each level image in parallel, and a multi-scale basic geometric feature point set is generated;

[0043] Based on the multi-scale basic geometric feature point set, a preset geometric constraint condition for the marking task is called to determine the constraint compliance of each feature point in the multi-scale basic geometric feature point set, to obtain a feature point subset meeting the constraint, and based on the feature point subset meeting the constraint, grouping association is performed to establish a candidate calibration marker structure set;

[0044] Based on the candidate calibration marker structure set, the feature points in each structure are traversed, the response intensity and the coordinate stability measurement value across the scale level in the multi-scale basic geometric feature point set are quantitatively detected, the contribution weight parameter of each feature point is determined, the weighted feature point contribution value is obtained, and based on the weighted feature point contribution value, the feature point coordinates associated with each candidate structure are fitted to obtain calibration reference point coordinates;

[0045] Based on the calibration reference point coordinates as the target position, the marking robot is instructed to perform a test marking, and the splashing and scattering pattern image of the marking robot nozzle tip at the moment of contact with the working ground is synchronously collected to obtain the micro marker pattern geometric feature, and based on the micro marker pattern geometric feature, the current position parameter of the tool tip of the marking robot is adjusted.

[0046] The acquisition steps of the multi-scale basic geometric feature point set are as follows:

[0047] The working site image data is subjected to multi-level resolution reconstruction, and a Gaussian pyramid generation method is used to adjust the resolution layer by layer and superimposed to the set to form a multi-resolution image set;

[0048] Based on the multi-resolution image set, corner point detection and line segment detection are performed on each level image in parallel, wherein the corner point detection calculates the corner point response value of each pixel based on the gradient intensity matrix, the line segment detection extracts line segment endpoints through edge direction histogram statistics and line segment connectivity criteria, and the corner points and line segment endpoints with response values exceeding a dynamic threshold are selected as candidate geometric response points to generate a multi-level geometric response point set;

[0049] The coordinate interpolation and scale attribute calculation are performed on each candidate geometric response point in the multi-level geometric response point set, wherein the coordinate interpolation adopts a quadratic polynomial fitting method to optimize the coordinate precision, and the scale attribute is determined based on the ratio of the level resolution and the original resolution where the geometric response point is located, to obtain the multi-scale basic geometric feature point set.

[0050] Specifically, for the work site image data named "Workshop Ground A Area" with resolution set to 1920x1080 pixels, a multi-level resolution reconstruction is performed, and a Gaussian pyramid generation process is started. First, the original image (0th layer) is subjected to Gaussian blur processing using a 5x5 Gaussian kernel with a sigma value of 1.0. Then, the blurred image is subjected to interlaced and columnar sampling to obtain a 1st layer image with a resolution of 960x540 pixels. The blur and sampling process is repeated to generate a 2nd layer image with a resolution of 480x270 pixels based on the 1st layer image. The three layers of images, i.e., the original image (1920x1080), the 1st layer image (960x540), and the 2nd layer image (480x270), are superimposed and stored to form a multi-resolution image set containing [Image Layer 0, Image Layer 1, Image Layer 2]. Subsequently, based on the multi-resolution image set, corner detection and line segment detection tasks are simultaneously started on the 0th, 1st, and 2nd layer images. For corner detection, taking a pixel point P(100, 150) on the 1st layer image as an example, the x-direction gradient I x and the y-direction gradient I y of all pixels in the 3x3 neighborhood of the pixel point P(100, 150) are calculated to obtain a gradient value matrix, and a gradient intensity matrix M is constructed For example, the gradient intensity matrix M is calculated as The parameter k of the Harris corner point response is set to 0.04, and the corner point response value R of the pixel is calculated as R = det(M) - k·trace(M) 2 = (850x600 - 120 2 )-0.04x(850+600) 2 = (510000 - 14400) - 0.04x(1450) 2 =

[0051] 495600 - 0.04x2102500 = 495600 - 84100 = 411500. The calculation is repeated for all pixels in the layer. For line segment detection, an edge detection algorithm such as the Canny operator is first applied to extract the edge points and their gradient directions in the image. Then, the edge image is analyzed to count the gradient direction histogram of the edge points in the local region, find the straight line direction corresponding to the histogram peak value, and connect the edge points based on the distance and direction consistency to form candidate line segments. The two end point coordinates of the line segments are extracted, for example, a line segment is detected in the 1st layer image with end points L1_start(50, 80) and L1_end(250, 85). Next, all calculated corner point response values and extracted line segment end points are screened, and a dynamic threshold is set based on the statistical characteristics of the response values of each layer of images. For example, for the corner point response values of the 1st layer image, the average value μ R and the standard deviation σR For example, μ R = 50000, σ R = 100000, set threshold T R = μ R + 1.5 × σ R = 50000 + 1.5 × 100000 = 200000, compare the corner response value R = 411500 with the threshold T R = 200000, 411500 > 200000, so the pixel point P(100, 150) is retained, and all the corner points and line segment endpoints screened through the respective thresholds jointly constitute the candidate geometric response points, all the points screened after the respective levels are collected to generate a multi-level geometric response point set containing [level 0 response point set, level 1 response point set, level 2 response point set], and then, for each candidate geometric response point in the multi-level geometric response point set, for example, the corner point P(100, 150) screened at the first level, the response value of which is 411500, the response values in the 3 × 3 neighborhood thereof are fitted by a quadratic polynomial surface, the sub-pixel coordinates of the peak of the surface are found by solving the partial derivatives, for example, the optimized coordinates are P'(100.21, 150.35), and then the scale attribute thereof is calculated, the point is located at the first level, the resolution thereof is 960 × 540, the original resolution is 1920 × 1080, so the scale attribute s = 960 / 1920 = 0.5, and the optimized coordinates (100.21, 150.35) and the scale attribute 0.5 are assigned to the point, and the coordinate interpolation and scale attribute calculation are performed on all the candidate geometric response points to obtain a multi-scale basic geometric feature point set containing accurate coordinates and scale information.

[0052] The acquisition step of the feature point subset satisfying the constraint is:

[0053] Based on the preset line marking task geometric constraint condition parameter configuration table, the minimum angle and the maximum angle of the target angle range, the lower threshold of the number of collinear points, and the minimum distance and the maximum distance of the point pair distance are called;

[0054] Each feature point in the multi-scale basic geometric feature point set is traversed, the angle formed by the feature point and the line segment connecting the adjacent two points is calculated, the number of collinear points continuously distributed on the line segment where the feature point is located is counted, the Euclidean distance between the feature point and the nearest neighbor point is measured, and it is respectively judged whether the angle formed by the feature point and the line segment connecting the adjacent two points is within the minimum angle and the maximum angle interval, whether the number of collinear points is greater than the lower threshold of the number of collinear points, and whether the Euclidean distance is within the minimum distance and the maximum distance interval of the point pair distance, and if any one of the judgment results is true, the feature point is marked as a valid candidate point, and a constraint compliance judgment result set is generated;

[0055] Based on the constraint compliance determination result set, all feature points marked as valid candidate points are filtered out, and unmarked feature points are removed to form a subset of feature points that satisfy the constraints.

[0056] Specifically, based on the preset geometric constraint parameter configuration table for the line drawing task, the minimum angle value of the target angle range is set to 178 degrees, and the maximum angle value is set to 182 degrees. This range is used to identify the point arrangement of approximate straight lines. The lower limit threshold for the number of collinear points is set to 5 points, requiring that a valid straight line segment contains at least 5 feature points. The minimum distance between point pairs is set to 8 pixels, and the maximum distance is set to 120 pixels, which is used to control the spatial distribution between valid feature points and their nearest neighbors. Then, the system traverses each feature point recorded in the multi-scale basic geometric feature point set generated in the previous step. Let the currently traversed feature point be P. i Its coordinates are (X i ,Y i ), with a scale of S i For angle constraints, find two adjacent feature points P within its neighborhood (e.g., at the same scale level or on the same detected line segment). i-1 and P i+1 Calculate vector and Calculate the angle between two vectors using the dot product. For example, if P i-1 (90.1, 148.2), P i (100.2, 150.3),

[0057] P i+1 (110.5, 152.4), then Calculate the included angle θ i ≈179.5 degrees. Determine if this angle is within the range [178, 182]. If it is (178≤179.5≤182), then the angle constraint determination result is true. Regarding the constraint on the number of collinear points, identify feature point P. i For the given line segment, count the total number N of continuously distributed feature points on that line segment. i For example, finding P i There are 7 feature points on the line segment L2. We determine if this number is greater than the lower limit threshold of 5 for the number of collinear points. If it is (7>5), the collinear point number constraint is true. For the point-to-point spacing constraint, we find the distance feature point P. i The nearest neighbor feature point P neighbor (Regardless of whether they are on the same straight line), calculate the Euclidean distance between the two points, for example, find the nearest neighbor P. neighbor(105.0,158.0), calculate the distance, determine whether the distance is in the [8, 120] pixel interval, and then the point pair distance constraint judgment result is true. Since P i The angle, the number of collinear points, and the point pair distance are all true, so the feature point P i is marked as an effective candidate point. The above calculation and judgment process is repeated for all feature points in the multi-scale basic geometric feature point set to generate a constraint compliance judgment result set that records whether each feature point is an effective candidate point, forming a feature point subset that satisfies the preset geometric constraint conditions.

[0058] The acquisition step of the candidate calibration marker structure set is:

[0059] Based on the preset corner point combination rule parameter table, the minimum allowed distance value between corner points, the maximum allowed distance value between corner points, the minimum allowed included angle value between corner points, the maximum allowed included angle value between corner points, the minimum allowed distance value between collinear points, the minimum required number of collinear points, and the fixed distance pairing allowed error value are called to construct a geometric grouping rule parameter set including the corner point combination distance interval, the corner point combination included angle range, the collinear arrangement distance threshold and the number threshold, and the fixed distance pairing error threshold;

[0060] All feature points in the feature point subset that satisfy the constraint are traversed, and the following operations are performed for each feature point: for the corner point combination rule, the straight line distance value between the feature point and other feature points is calculated, and it is determined whether the distance value is in the corner point combination distance interval. At the same time, the included angle value of the connecting line of the two points with the reference coordinate axis is calculated, and it is determined whether the angle value is in the preset included angle range. The feature point pairs that satisfy both the distance interval and the angle range conditions are selected as candidate corner point combinations. For the collinear arrangement rule, the distance value of adjacent feature points on the straight line segment where the feature point is located is searched, the number of feature points with continuous distance not exceeding the collinear arrangement distance threshold is counted, and the feature point sequence whose number reaches the minimum required number of collinear points is selected as a candidate collinear group. For the fixed distance pairing rule, the actual distance value between the feature point and all feature points is measured, and the feature point pairs whose absolute value of the difference between the actual distance and the preset fixed distance does not exceed the allowed error value are selected as candidate fixed distance pairing groups, to generate a candidate grouping set containing candidate corner point combinations, candidate collinear groups, and candidate fixed distance pairing groups.

[0061] Based on the candidate grouping set, in the case that the same feature point is assigned to multiple candidate groups, the group that meets the most grouping rules is preferentially retained, the groups that have coordinate overlap or logical conflicts with other groups are eliminated, and the remaining independent and conflict-free candidate corner point combinations, candidate collinear groups, and candidate fixed distance pairing groups are combined into complete feature point combinations to establish a candidate calibration marker structure set.

[0062] Specifically, based on the preset corner point combination rule parameter table, the system calls a series of parameter values for geometric grouping, specifically including: a minimum allowed distance value between corner points, set to 40 pixels, a maximum allowed distance value between corner points, set to 200 pixels, a minimum allowed included angle value between corner points, set to 85 degrees, a maximum allowed included angle value between corner points, set to 95 degrees, a maximum allowed spacing value between collinear points, set to 15 pixels, a minimum number of collinear points, set to 4 points, and a fixed distance pairing allowed error value, set to 3 pixels. These parameters jointly construct a geometric grouping rule parameter set including a corner point combination distance interval [40, 200] pixels, a corner point combination included angle range [85, 95] degrees, a collinear arrangement spacing threshold of 15 pixels and a number threshold of 4 points, and a fixed distance pairing error threshold of 3 pixels. Subsequently, all feature points in the feature point subset generated in the previous step that satisfy the constraints are traversed, and a geometric grouping rule matching operation is performed on each feature point P. For the corner point combination rule, the straight line distance d(P, Q) between P and all other feature points Q is calculated, and it is determined whether d(P, Q) is within the [40, 200] pixel interval. If it is satisfied, a third feature point R is further sought, the angle value of the angle ∠QPR (or ∠PQR, etc., depending on the rule definition, and ∠QPR is taken as an example here) formed by P, Q, and R is calculated, and it is determined whether the angle value is within the [85, 95] degree range. For example, for P(100.2, 150.3), Q(102.5, 248.1), and R(195.8, 155.0), d(P, Q) ≈ 98 pixels (within [40, 200]), d(P, R) ≈ 96 pixels (within [40, 200]), and the included angle ∠QPR ≈ 91.5 degrees (within [85, 95]). Therefore, P, Q, and R form a candidate corner point combination {P, Q, R}. For the collinear arrangement rule, adjacent feature points on the straight line segment where the feature point P is located are searched, and the spacing values between them are calculated. For example, the ordered points on the line segment L2 where P is located are P1, P2, P, P4, and P5, the spacing d(P1, P2) = 12, d(P2, P) = 10, d(P, P4) = 13, and d(P4, P5) = 11. All spacings do not exceed the collinear arrangement spacing threshold of 15 pixels, and the number of feature points in the continuous sequence is 5. It is determined whether the number reaches the minimum number of collinear points, which is 4. Yes (5 ≥ 4), and the feature point sequence {P1, P2, P, P4, P5} is selected as a candidate collinear group. For the fixed distance pairing rule, the actual distance d(P, Q) between the feature point P and all other feature points Q is calculated, and it is determined whether the absolute value of the difference between the distance and the preset fixed distance of 150 pixels does not exceed the allowed error value of 3 pixels, i.e., |d(P, Q) - 150| ≤ 3. For example, the distance between P and Q ′ (248.5, 160.9) is calculated to be 150.2 pixels, and |150.2 - 150| = 0.2 ≤ 3. Therefore, P and Q form a candidate fixed distance pairing {P, Q}. Pixel, the absolute value of the difference is calculated |148.68-150|=1.32, and it is judged that 1.32≤3, yes, so the feature point pair {P, Q'} is screened out as a candidate fixed distance pairing group. After the above rule matching is completed for all feature points, a candidate grouping set containing all found candidate corner point combinations, candidate collinear groups and candidate fixed distance pairing groups is generated. Finally, potential allocation conflicts are processed, and it is checked whether there is a feature point that is allocated to multiple candidate groups, for example, point P belongs to candidate corner point combination {P, Q, R} and candidate collinear group {P1, P2, P, P4, P5} at the same time. The group with the most members is retained to form a complete feature point combination set, which is the candidate calibration marker structure set.

[0063] The acquisition step of the weighted feature point contribution value is:

[0064] Each feature point in the candidate calibration marker structure set is traversed to extract the response intensity of the feature point in the corner point detection of each level image and the coordinate data set of the feature point at different resolution levels in the multi-scale basic geometric feature point set, and a feature point response intensity-multi-scale coordinate data set is generated.

[0065] Based on the feature point response intensity-multi-scale coordinate data set, the coordinate stability measurement value of each feature point across the scale level is calculated to obtain a feature point coordinate stability set, and the formula is:

[0066]

[0067] Wherein, S j is the coordinate stability measurement value of feature point j, k is the number of levels of feature point j across the scale level in the multi-scale basic geometric feature point set, x jm is the x-axis coordinate value of feature point j in the mth level image, y jm is the y-axis coordinate value of feature point j in the mth level image, is the x-axis coordinate mean value of feature point j, is the y-axis coordinate mean value of feature point j;

[0068] Based on the feature point response intensity-multi-scale coordinate data set and the feature point coordinate stability set, the weighted feature point contribution value is calculated, and the formula is:

[0069]

[0070] Wherein, C p is the weighted feature point contribution value of feature point p, R p is the response intensity of feature point p, S p is the coordinate stability measurement value of feature point p, max(R q) is the maximum response intensity among all feature points, max(S q ) is the maximum coordinate stability metric value among all feature points.

[0071] Specifically, for each included feature point p in the candidate calibration marker structure set established in the previous step, the following operations are performed to calculate its weighted contribution value. First, the response intensity values corresponding to the feature point p on the respective level images on which it was detected during the initial corner detection are extracted from the multi-scale base geometric feature point set, for example, the response value of the feature point p (i.e. point 3 before) on the 0th level is R p,0 = 650, the response value on the 1st level is R p,1 = 680, and the response value on the 2nd level is R p,2 = 590. Generally, the response value on the most stable or highest resolution level is taken, or the average value / maximum value is taken as the representative response intensity R p of the feature point p, where the response value on the highest level, i.e. the 0th level, is taken as R p = R p,0 = 650. At the same time, the coordinate data set recorded in the multi-scale base geometric feature point set across different resolution levels of the feature point p is extracted, i.e. its (interpolated optimized) coordinates (x pm , y pm ) on each level on which it was detected, where m represents the level index, as in the previous example (coordinates have been normalized to the original resolution): the coordinate set of p = 3 is {level 1 (m = 1): (100.35, 200.72), level 2 (m = 2): (100.24, 200.60), level 3 (m = 3): (100.00, 200.40)} (here the level index starts from 1, and there are k = 3 levels). These extracted response intensity and multi-scale coordinate data are integrated to generate the response intensity-multi-scale coordinate data set of the feature point p. Then, based on the multi-scale coordinate data in the data set, the cross-scale level coordinate stability metric value S p of the feature point p is calculated, and the calculation formula is as follows:

[0072]

[0073] Formula explanation: S j represents the coordinate stability metric value of feature point j, and the calculation result is between 0 and 1. The closer the value is to 1, the higher the stability is, k is the number of scale levels in which the feature point j appears and whose coordinates are recorded, indicates the summation of the k levels, x jm and y jm are the (normalized) x-axis and y-axis coordinate values of the feature point j in the mth scale level image, and are the mean values of x-axis and y-axis coordinates of feature point j in k levels respectively, and the calculation method is and The core part of the formula is to calculate the root mean square deviation (RMSD) of coordinate point (x jm ,y jm ) relative to its average position , which reflects the dispersion or instability of the coordinate point at different scales. Finally, the instability measure (RMSD) is converted to stability measure S j by 1 / (1+RMSD). The smaller the RMSD (the more stable), the closer S j to 1, and vice versa. The larger the RMSD (the more unstable), the closer S j to 0. Add 1 to avoid zero denominator and ensure the value range.

[0074] Parameter acquisition and examples: for feature point p=3, k=3, the coordinate set (normalized) is {(100.35, 200.72), (100.24, 200.60), (100.00, 200.40)},

[0075] Calculate the mean coordinate:

[0076]

[0077] Calculate the square of RMSD:

[0078]

[0079] Calculate the RMSD:

[0080] Calculate the stability

[0081] Store the calculation results S j of all feature points in the feature point coordinate stability set. Then, based on the response intensity R p =650 and the coordinate stability S p =0.8355 of feature point p, as well as the maximum response intensity max(R q ) and the maximum coordinate stability max(S q ) calculated in all feature points (for example, max(R q )=820 and max(S q )=0.9600 by traversing all feature points), calculate the weighted feature point contribution value C p , and the calculation formula is:

[0082]

[0083] Formula explanation: C p The weighted feature point contribution value of feature point p is a normalized comprehensive evaluation score, R p is the response intensity of feature point p, S p is the coordinate stability measure value of feature point p, R p ·S p represents the original contribution combined with intensity and stability, max(R q ) is the maximum response intensity value observed in all candidate feature points, q traverses all feature points, max(S q ) is the maximum coordinate stability measure value calculated in all candidate feature points, and the product of the two maximum values is used as the denominator for normalization, which scales the value of C p to the approximate [0, 1] interval, so that the contribution value has comparability. The core logic of this formula is to reward feature points with both high response intensity (easy to detect) and high cross-scale coordinate stability (reliable positioning) through the product R p ·S p , and normalize it by the global maximum value.

[0084] For feature point Repeat this calculation for each feature point in the candidate calibration marker structure set to obtain the weighted contribution value of all feature points.

[0085] The benefit of the formula is that S j quantifies the coordinate consistency of feature points under different resolution images, improving the recognition ability of feature points that are not sensitive to scale changes and have high positioning accuracy, C p then considers the detection significance (R p ) and positioning stability (S p ) of feature points, and provides a relatively objective quantitative index that measures the importance of each feature point to subsequent calibration tasks through normalization. Compared with using R p or geometric properties alone, it can better filter out high-quality feature points.

[0086] The result C3≈0.6898 indicates that feature point 3 has a moderately high contribution value, and its response intensity and stability are relatively reliable in combination. This contribution value will be used as a weight in the next step of calculating the calibration reference point coordinates.

[0087] The steps for obtaining the calibration reference point coordinates are:

[0088] Traverse each candidate structure in the candidate calibration marker structure set, extract the weighted feature point contribution value set and the corresponding feature point coordinate set of all feature points in the structure, and generate a candidate structure feature point contribution-coordinate data set;

[0089] Calculate the fitting coordinates of the calibration reference point based on the candidate structure feature point contribution-coordinate dataset Generate the calibration reference point coordinates of all candidate structures, the formula is:

[0090] And

[0091] Where C i is the weighted feature point contribution value of the i-th feature point in the candidate structure, X i is the x-axis coordinate value of the i-th feature point, Y i is the y-axis coordinate value of the i-th feature point, and n is the total number of feature points in the candidate structure.

[0092] Specifically, for each candidate structure g in the candidate calibration marker structure set generated in the previous step, the system first extracts all feature points contained in the structure g. For each feature point i, it obtains its corresponding weighted feature point contribution value C i and its accurate coordinates (X i , Y i ) from the calculation results of the previous step, collects these data, and generates a feature point contribution-coordinate dataset for the candidate structure g. For example, the dataset for structure g is: {Point 1: C1=0.650, (X1,Y1)=(90.1,100.5); Point 3: C3=0.6898, (X3,Y3)=(100.2,150.3); Point 5: C5=0.580, (X5,Y5)=(110.8,99.8)}, which contains n=3 feature points in total. Then, based on this dataset, the weighted average formula is applied to calculate the fitting coordinates of the calibration reference point of the candidate structure g The calculation formulas are respectively:

[0093]

[0094] Formula explanation: and represent the x-axis and y-axis coordinates of the fitting calibration reference point of the candidate structure g, and n is the total number of feature points in the structure, indicates the summation of all n feature points in the structure, C i is the weighted feature point contribution value of the i-th feature point (as the weight), X i and Y i are the x-axis and y-axis coordinate values of the i-th feature point, the numerator calculates the weighted sum of all feature point x coordinates, the numerator calculates the weighted sum of all feature point y coordinates, and the denominator The sum of all feature point weights is calculated, and the entire formula calculates the weighted average of the coordinates of the feature points in the structure. The higher the contribution value of a feature point, the greater the impact of the final reference point position. This calculation aims to obtain a more stable and accurate reference point coordinate that represents the center or key position of the geometric structure by giving higher weights to high-quality feature points.

[0095] Parameter acquisition and examples: using the data set of structure g for calculation,

[0096] Calculate the sum of weights:

[0097] Calculate the weighted sum of x coordinates:

[0098] Calculate the weighted sum of y coordinates:

[0099] Calculate the fitting coordinates:

[0100]

[0101] Therefore, the fitted coordinates of the calibration reference point of the candidate structure g are (99.98, 118.18). This calculation is performed for each structure in the candidate calibration marker structure set, generating a list of calibration reference point coordinates for all candidate structures.

[0102] The advantage of the formula is that by using the weighted feature point contribution value C i The coordinates are averaged as weights rather than simple arithmetic mean (geometric center), making the calculated calibration reference point position more resistant to positioning errors caused by individual low-quality (low response strength or low stability) feature points in the structure, improving the robustness and accuracy of the reference point coordinates.

[0103] The result represents the best representative point position of the geometric structure determined by considering the quality of each feature point in the structure. This coordinate will be used as the target point for the robot to perform the test marking.

[0104] The acquisition steps of the geometric characteristics of the micro-marking pattern are:

[0105] The calibration reference point coordinates are transmitted to the marking robot console to trigger the robot to perform the test marking operation, and the splashing and scattering pattern image of the nozzle tip at the moment of contact with the ground is synchronously collected to generate a splashing image sequence.

[0106] Based on the splash image sequence, the following processing is performed on each frame image: the gray threshold value is determined by the maximum inter-class variance method to complete the binary segmentation, the morphological closing operation is performed on the binary image to eliminate the holes, the Canny operator is applied to extract the splash area edge, and the splash contour data set is generated;

[0107] The splash contour data set is traversed, and the geometric moment calculation is performed on each closed contour point set: the first moment M 10 and the first moment M 01 in the y-axis direction of the closed contour point set in the x-axis direction are calculated; 00 The centroid physical coordinates x c and y 10 are obtained by x 00 =M c / M 01 and y 00 =M c / M c , and the micro-mark shape geometric feature is constructed.

[0108] Specifically, one or more calibration reference point coordinates calculated in the previous step, such as the coordinates of the structure g are transmitted to the control console system of the marking robot. After the control system receives the target coordinates, it drives the robot to move, so that the nozzle end moves above the corresponding physical position, and then triggers a short spraying action to perform a test marking operation. At the moment when the robot nozzle end touches the ground (or very close to the ground) and sprays paint, the camera associated with the nozzle (such as the camera installed on the robot end effector, covering the area below the nozzle in the field of view) is started to capture images synchronously to obtain the image of the splash dispersion shape. A single frame image or a short image sequence may be captured to form a splash image sequence. Then, image processing procedures are performed on each frame of valid image (records the image of the splash shape, such as the frame after the splash stabilizes) in this splash image sequence. First, the maximum inter-class variance method is applied to the gray scale image to automatically calculate an optimal gray scale threshold T otsu to distinguish the splash area (foreground) and the background. For example, for a 200x200 pixel splash image, T otsu= 128, and then binarizing the image according to the threshold value, setting the pixel gray value greater than 128 as 255 (white, representing the splashing area) and less than or equal to 128 as 0 (black, representing the background), to obtain a binary image, and then performing a morphological closing operation on the binary image, selecting a circular or square structural element with a size of, for example, 3x3 pixels, performing an expansion operation on the white area first, and then performing a corrosion operation, which can effectively fill small holes or breaks that may exist in the splashing area while having little effect on the edge, keeping the splashing main body shape complete, after processing, applying a Canny edge detection operator to extract the edge contour of the splashing area in the processed binary image, performing edge connection to generate a set of pixel point coordinates constituting the closed contour of the splashing area, for example, obtaining a contour point set containing N contour points, collecting the contour data obtained from all processing frames to form a splashing contour data set, and finally, traversing each closed contour point set in the splashing contour data set to perform geometric moment calculation on it, calculating the zeroth moment M 00 (area, number of pixels) of the region surrounded by the contour, the first moment M 10 in the x-axis direction, and the first moment M 01 in the y-axis direction, for example, calculating the area of the splashing area M 00 = 950 pixels 2 , the x-direction first moment M 10 = 94850 pixels 3 , and the y-direction first moment M 01 = 112200 pixels 3 , taking the zeroth moment M 00 as the reference, calculating the centroid physical coordinates = M 10 / M 00 = 94850 / 950 = 99.84 pixels, y c = M 01 / M 00 = 112200 / 950 = 118.11 pixels, taking the calculated centroid coordinates (x c , y c ) = (99.84, 118.11) as the core geometric feature describing the morphology of the micro-marking of this test scribing, and constructing the micro-marking morphology geometric feature data containing the centroid coordinates.

[0109] Based on the micro-marking morphology geometric feature, the spatial deviation vector between the centroid position coordinates and the calibration reference point coordinates is calculated, and the steps of adjusting the current position parameter and attitude parameter setting of the tool end point of the scribing robot according to the spatial deviation vector are as follows:

[0110] Based on the centroid physical coordinates of the micro-mark shape geometric features, the difference between the centroid physical coordinates and the preset coordinate components is calculated axis by axis to generate a deviation vector dataset including x-axis deviation and y-axis deviation;

[0111] According to the deviation vector dataset, the deviation is decomposed into a lateral translation compensation step and a longitudinal translation compensation step in the robot base coordinate system to generate a pose correction parameter set;

[0112] Based on the pose correction parameter set, the current position parameters of the tool end point of the line marking robot are adjusted.

[0113] Specifically, based on the micro-mark shape geometric features obtained in the previous step, i.e. the centroid physical coordinates (x c ,y c ) of the actual splashing mark (99.84, 118.11), the system starts to calculate the spatial deviation vector between the actual mark position and the expected calibration reference point coordinates First, the difference between the coordinate components is calculated axis by axis, the x-axis deviation is calculated in units of millimeters, the y-axis deviation is calculated in units of millimeters, and the two deviation amounts are combined into a deviation vector and the vector is stored in the deviation vector dataset. The calculated deviation vector is decomposed into a compensation movement amount in the robot base coordinate system.

[0114] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can modify or change the above disclosed technical content to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made according to the technical essence of the present application to the above embodiments without departing from the technical solution content of the present application still falls within the protection scope of the present application.

Claims

1. A vision-guided automatic calibration method for a line-marking robot, characterized in that, Includes the following steps: Based on the image data of the work site acquired by the line marking robot, an image set including different resolution levels is established. Corner detection and line segment detection are performed in parallel on each level of image to generate a multi-scale basic geometric feature point set. Based on the multi-scale basic geometric feature point set, the preset geometric constraints for the line drawing task are invoked to determine the constraint compliance of each feature point in the multi-scale basic geometric feature point set, thereby obtaining a subset of feature points that satisfy the constraints. Based on the subset of feature points that satisfy the constraints, grouping and associating them, a candidate calibration mark structure set is established. Based on the candidate calibration marker structure set, the feature points within each structure are traversed, the detection response intensity and the coordinate stability metric value across scale levels in the multi-scale basic geometric feature point set are quantified, the contribution weight parameter of each feature point is determined, and the weighted feature point contribution value is obtained. Based on the weighted feature point contribution value, the coordinates of the feature points associated with each candidate structure are fitted to obtain the coordinates of the calibration reference point. Based on the coordinates of the calibration reference point as the target position, the marking robot is instructed to perform trial marking. Simultaneously, an image of the spray pattern at the moment the nozzle tip of the marking robot contacts the working ground is acquired to obtain the geometric features of the micro-marker morphology. Based on the geometric features of the micro-marker morphology, the current position parameters of the tool end point of the marking robot are adjusted.

2. The vision-guided automatic calibration method for a line-marking robot according to claim 1, characterized in that, The steps for obtaining the multi-scale basic geometric feature point set are as follows: Multi-level resolution reconstruction is performed on the on-site image data. The resolution is adjusted layer by layer and superimposed into the set to form a multi-resolution image set. Based on the multi-resolution image set, corner detection and line segment detection are performed in parallel on each level of image. Corner detection calculates the corner response value of each pixel based on the gradient intensity matrix. Line segment detection extracts the endpoints of line segments through edge direction histogram statistics and line segment connectivity criteria. Corner points and line segment endpoints with response values ​​exceeding the dynamic threshold are selected as candidate geometric response points to generate a multi-level geometric response point set. For each candidate geometric response point in the multi-level geometric response point set, coordinate interpolation and scale attribute calculation are performed. The coordinate interpolation adopts a quadratic polynomial fitting method to optimize the coordinate accuracy. The scale attribute is determined based on the ratio of the resolution of the level where the geometric response point is located to the original resolution, thus obtaining a multi-scale basic geometric feature point set.

3. The vision-guided automatic calibration method for a line-marking robot according to claim 1, characterized in that, The steps for obtaining the feature point subset that satisfies the constraints are as follows: Based on the preset geometric constraint parameter configuration table for the line drawing task, the minimum and maximum angles of the target angle range, the lower limit threshold of the number of collinear points, and the minimum and maximum distances between point pairs are called. Traverse each feature point in the multi-scale basic geometric feature point set, calculate the angle formed by the line connecting the feature point and two adjacent points, count the number of collinear points continuously distributed on the line segment where the feature point is located, measure the Euclidean distance between the feature point and its nearest neighbor, and determine whether the angle formed by the line connecting the feature point and two adjacent points is within the range of minimum and maximum angles, whether the number of collinear points is greater than the lower limit threshold of the number of collinear points, and whether the Euclidean distance is within the range of minimum and maximum distance between point pairs. If any of the judgment results is true, mark the feature point as a valid candidate point and generate a constraint compliance judgment result set. Based on the constraint compliance determination result set, all feature points marked as valid candidate points are filtered out, and unmarked feature points are removed to form a subset of feature points that satisfy the constraints.

4. The vision-guided automatic calibration method for a line-marking robot according to claim 1, characterized in that, The steps for obtaining the candidate labeling structure set are as follows: Based on the preset corner point combination rule parameter table, the minimum allowable distance between corner points, the maximum allowable distance between corner points, the minimum allowable included angle between corner points, the maximum allowable included angle between corner points, the minimum allowable spacing between collinear points, the minimum requirement for the number of collinear points, and the allowable error value for fixed distance pairing are called to construct a geometric grouping rule parameter set including the corner point combination distance interval, the corner point combination included angle range, the collinear arrangement spacing threshold and quantity threshold, and the fixed distance pairing error threshold. Iterate through all feature points in the subset of feature points that satisfy the constraints, and perform the following operations for each feature point: For the corner point combination rule, calculate the straight-line distance between the feature point and other feature points, determine whether the distance value is within the corner point combination distance interval, and simultaneously calculate the angle between the line connecting the two points and the reference coordinate axis, determine whether the angle value is within the preset angle range, and select feature point pairs that simultaneously satisfy the distance interval and angle range conditions as candidate corner point combinations; For the collinear arrangement rule, search for the distance values ​​between adjacent feature points on the straight line segment where the feature point is located, count the number of feature points whose continuous distance does not exceed the collinear arrangement distance threshold, and select feature point sequences whose number reaches the minimum requirement for the number of collinear points as candidate collinear groups; For the fixed distance pairing rule, measure the actual distance values ​​between the feature point and all feature points, and select feature point pairs whose absolute value of the difference between the actual distance and the preset fixed distance does not exceed the allowable error value as candidate fixed distance pairing groups, generating a candidate grouping set containing candidate corner point combinations, candidate collinear groups, and candidate fixed distance pairing groups; Based on the candidate group set, when the same feature point is assigned to multiple candidate groups, the group that meets the most grouping rules is retained first, and the group that has coordinate overlap or logical conflict with other groups is eliminated. The remaining independent and conflict-free candidate corner point combination, candidate collinear group, and candidate fixed distance pairing are combined into a complete feature point combination to establish a candidate calibration mark structure set.

5. The vision-guided automatic calibration method for a line-marking robot according to claim 1, characterized in that, The steps for obtaining the weighted feature point contribution value are as follows: Traverse each feature point in the candidate calibration marker structure set, extract the response intensity of the feature point in corner detection of the image at each level, and the coordinate data set at different resolution levels in the multi-scale basic geometric feature point set to generate the feature point response intensity-multi-scale coordinate dataset. Based on the feature point response intensity-multi-scale coordinate dataset, the coordinate stability metric of each feature point across scale levels is calculated to obtain the feature point coordinate stability set. Based on the feature point response intensity-multiscale coordinate dataset and the feature point coordinate stability set, the weighted feature point contribution value is calculated.

6. The vision-guided automatic calibration method for a line-marking robot according to claim 1, characterized in that, The steps for obtaining the coordinates of the calibration reference point are as follows: Traverse each candidate structure in the candidate calibration label structure set, extract the weighted feature point contribution value set and the corresponding feature point coordinate set of all feature points in the structure, and generate the candidate structure feature point contribution-coordinate dataset; Based on the candidate structure feature point contribution-coordinate dataset, the calibration reference point fitting coordinates are calculated, and the calibration reference point coordinates of all candidate structures are generated.

7. The vision-guided automatic calibration method for a line-marking robot according to claim 1, characterized in that, The steps for obtaining the geometric features of the micro-marker morphology are as follows: The coordinates of the calibration reference point are transmitted to the control console of the line drawing robot, triggering the robot to perform trial line drawing operations. Simultaneously, images of the splash dispersion pattern at the moment when the nozzle tip contacts the ground are acquired, generating a splash image sequence. Based on the splash image sequence, the following processing is performed on each frame: the grayscale threshold is determined by the maximum inter-class variance method to complete the binarization segmentation, morphological closing operation is performed on the binary image to eliminate holes, the Canny operator is applied to extract the edge of the splash area, and a splash contour dataset is generated. Iterate through the splatter profile dataset and perform geometric moment calculations for each closed profile point set: calculate the first moment M of the closed profile point set in the x-axis direction. 10 and the first moment M in the y-axis direction 01 With the zero-order moment M of the contour region 00 Based on x c =M 10 / M 00 With y c =M 01 / M 00 Obtain the physical coordinates x of the centroid c and y c Construct the geometric features of micro-marker morphology.

8. The vision-guided automatic calibration method for a line-marking robot according to claim 1, characterized in that, Based on the geometric features of the micro-marker morphology, the spatial deviation vector between the centroid position coordinates and the calibration reference point coordinates is calculated. The steps for adjusting the current position and attitude parameters of the tool end point of the marking robot according to this spatial deviation vector are as follows: Based on the centroid physical coordinates of the micro-marker morphology geometric features, the difference between the centroid physical coordinates and the preset coordinate components is calculated axis by axis to generate a deviation vector dataset including x-axis deviation and y-axis deviation. Based on the deviation vector dataset, the deviation is decomposed into the lateral translation compensation step size and the longitudinal translation compensation step size in the robot base coordinate system, and a set of attitude correction parameters is generated. Based on the set of posture correction parameters, the current position parameters of the tool end point of the marking robot are adjusted.

Images