A 3D contour-based sole glue spraying track generation method and system

Abstract

A kind of shoe sole glue spraying track generation method based on 3D profile, comprising the following steps: obtaining point cloud data containing shoe sole;The point cloud data is segmented into foreground point cloud and background point cloud, and the shoe sole point cloud is foreground point cloud;Determine the main direction of shoe sole point cloud, adjust the main direction of shoe sole point cloud to the same direction as the flow line;Resample shoe sole point cloud to obtain resampled secondary shoe sole point cloud;The secondary shoe sole point cloud is denoised;The denoised secondary shoe sole point cloud is divided into toe segment point cloud, midfoot segment point cloud and heel segment point cloud according to the direction from shoe sole heel to shoe sole tip;Search glue spraying point position, search toe segment point cloud, midfoot segment point cloud and heel segment point cloud respectively, obtain toe glue spraying point position of toe segment point cloud, midfoot glue spraying point position of midfoot segment point cloud and heel segment glue spraying point position of heel segment point cloud.The present application intelligently generates shoe sole glue spraying track, provides track data for mechanical hand glue spraying, without manual intervention, improves production efficiency.

Description

Technical Field

[0001] This invention belongs to the field of shoe sole adhesive spraying technology, specifically a method and system for generating shoe sole adhesive spraying trajectory based on 3D contours. Background Technology

[0002] Industrial automation equipment is now widespread in the footwear manufacturing industry, and the level of intelligence in these devices is gradually improving. In particular, the introduction of machine vision has laid the foundation for the intelligent production of irregularly shaped products like shoes, and machine vision-guided robotic flexible glue spraying operations are also emerging in the industry. Currently, there are two main methods for automated robotic glue spraying:

[0003] The first method involves manually teaching the robot and generating the glue spraying trajectory. This method involves programming the robot to generate the glue spraying trajectory for different styles and sizes of shoes. This process requires the participation of professionals and is quite complicated. It is highly dependent on the experience of the construction workers. In addition, due to the process of shoe production, there will be a certain range of deformation, which will result in the known trajectory not being accurate enough, and will ultimately affect the glue spraying effect.

[0004] The second method is to model the sole of the shoe based on machine vision 3D imaging technology, and use the sole model data to generate the glue path trajectory of the sole, thereby guiding the robot to spray glue; however, the current technology is generally not very adaptable, and requires professionals to perform post-processing, manual editing and optimization of the 3D model, etc., with a lot of human intervention and a low level of intelligence. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for generating adhesive spraying trajectory on shoe soles based on 3D contours.

[0006] A method for generating adhesive spraying trajectory on shoe soles based on 3D contours includes the following steps:

[0007] Obtain point cloud data including the shoe sole;

[0008] The point cloud data is segmented into foreground point cloud and background point cloud, with the shoe sole point cloud being the foreground point cloud.

[0009] Direction processing: Determine the main direction of the sole point cloud and adjust the main direction of the sole point cloud to be the same as the direction of the production line;

[0010] Resampling process: The shoe sole point cloud that has been oriented is resampled to obtain a resampled secondary shoe sole point cloud;

[0011] Denoising processing: Denoising processing is performed on the secondary shoe sole point cloud;

[0012] Segmented processing: The noise-reduced secondary sole point cloud is divided into toe segment point cloud, mid segment point cloud, and heel segment point cloud according to the direction from the heel to the toe of the sole.

[0013] Search for glue spraying points by searching the point clouds of the toe section, the mid-section of the shoe, and the heel section, respectively, to obtain the glue spraying points of the toe section point cloud, the mid-section of the shoe point cloud, and the heel section point cloud.

[0014] Connect the glue spraying points at the toe, mid-shoe, and heel in sequence to form a glue spraying trajectory.

[0015] The segmentation process employs a random sampling consensus algorithm.

[0016] As a further improvement, the direction processing is performed using principal component analysis. If the principal direction of the shoe sole point cloud is different from the flow line direction, the shoe sole point cloud is rotated to make the principal direction of the shoe sole point cloud the same as the flow line direction.

[0017] As a further improvement, the noise reduction process specifically includes:

[0018] Set the search radius SR and the threshold for the number of neighboring points NP. Traverse each point in the secondary shoe sole point cloud, search for points within the radius SR around the current point, and count the number of searched points TP. If TP is less than NP, then the current point is a noise point and the noise needs to be removed. After traversal, the denoised shoe sole point cloud is obtained.

[0019] As a further improvement, the point cloud of the shoe toe segment is searched using the following method:

[0020] There is a toe and mid-shoe edge line between the toe and mid-shoe sections. The midpoint of the toe and mid-shoe edge line is taken as the center TC. The center TC divides the toe and mid-shoe edge line into a starting ray TL0 on the left and a terminating ray TLn on the right. The position with the largest jump along the positive direction of the starting ray TL0 is the quasi-adhesive point TP_0. The point with a lower height than the quasi-adhesive point TP_0 and a height difference less than TH along the negative direction of the starting ray TL0 is the quasi-adhesive point TP_1. The quasi-adhesive point TP_1 is set as the adhesive point TP0. The starting ray TL0 is rotated counterclockwise around the center TC by a preset angle to form ray TL1. The search process along the starting ray TL0 is repeated to obtain the adhesive point TP1. The preset angle is rotated sequentially to the terminating ray TLn on the right side of the center TC to obtain all the adhesive point positions {TP0, TP1, ..., TPn} of the toe section point cloud.

[0021] Here, "jump" refers to the height difference between the found quasi-spray point and the center TC. "Positive direction" refers to the direction along the starting ray TLO from the center TC, and "negative direction" refers to the direction from the outside of the starting ray TL0 toward the center TC.

[0022] As a further improvement, the point cloud of the shoe heel segment is searched using the following method:

[0023] There is a heel and mid-shoe edge line between the heel and mid-shoe sections. The midpoint of the heel and mid-shoe edge line is taken as the center BC. The center BC divides the heel and mid-shoe edge line into a starting ray BL0 on the left and a terminating ray BLn on the right. The position with the largest jump along the positive direction of the starting ray BL0 is the quasi-adhesive point BP_0. The point with a height lower than BP_0 and a height difference less than BH along the opposite direction of the starting ray BL0 is the quasi-adhesive point BP_1. The quasi-adhesive point BP_1 is set as the adhesive point BP0. The starting ray BL0 is rotated clockwise around the center BC by a preset angle to form ray BL1. The search process along the starting ray BL0 is repeated to obtain the adhesive point BP1. The preset angle is rotated sequentially to the terminating ray BLn on the right side of the center BC to obtain all the adhesive point positions {BP0, BP1, ..., BPn} of the heel section point cloud.

[0024] The positive direction of the initial ray BL0 refers to the direction along the initial ray BLO starting from the center BC, while the negative direction of the initial ray BL0 refers to the direction from the outside of the initial ray BL0 toward the center BC.

[0025] As a further improvement, the point cloud of the midsection of the shoe is searched using the following method:

[0026] The starting line ML0 is the edge line between the toe and the middle of the shoe, and the ending line MLn is the edge line between the heel and the middle of the shoe. The midpoint of all point clouds on the starting line ML0 is the search starting point MB0. The position with the largest jump is found to the left along ML0 as the quasi-adhesive point MPL_0. The point with a height lower than MPL_0 and a height difference less than MH is searched to the right along ML0 as the quasi-adhesive point MPL_1. The quasi-adhesive point MPL_1 is set as the adhesive point MPL0.

[0027] Find the position with the largest jump to the right along ML0 as the quasi-adhesion point MPR_0. Search for the point with a height lower than MPR_0 and a height difference less than MH to the left along ML0 as the quasi-adhesion point MPR_1. Set the quasi-adhesion point MPR_1 as the adhesion point MPR0. Translate the starting line ML0 along its orthogonal direction toward the edge of the shoe heel by a preset distance to form line ML1. Repeat the search process along the starting line ML0 to obtain the adhesion point MPL1 on the left and the adhesion point MPR1 on the right. Translate the line by a preset distance to the termination line MLn to obtain all the adhesion point positions {MPL0, MPL1, ..., MPLn} and {MPR0, MPR1, ..., MPRn} of the mid-shoe point cloud.

[0028] A 3D contour-based shoe sole adhesive spraying trajectory generation system includes:

[0029] The acquisition unit is used to acquire point cloud data of the shoe sole;

[0030] Segmentation unit, used to segment point cloud data into foreground point cloud and background point cloud;

[0031] Rotation unit, used to rotate the foreground cloud;

[0032] Noise reduction unit, used for noise reduction of shoe sole dot cloud;

[0033] Segmentation units are used to divide the sole point cloud into toe segment point cloud, mid-shoe segment point cloud, and heel segment point cloud;

[0034] The search unit is used to search for glue spray points on the point cloud of the toe section, the point cloud of the mid section, and the point cloud of the heel section.

[0035] Compared with the prior art, the present invention has the following beneficial technical effects:

[0036] By employing a specific method for searching adhesive spraying points, accurate spraying locations can be obtained, and intelligent spraying trajectories for shoe soles can be generated. This provides trajectory data for robotic arms to spray adhesive without human intervention, thereby improving production efficiency. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0038] Figure 2 This is a flowchart of the algorithm for searching adhesive spray points on the toe section of the shoe according to the present invention;

[0039] Figure 3 This is a flowchart of the algorithm for searching adhesive spray points on the heel segment of the present invention. Detailed Implementation

[0040] To further understand the features, technical means, and specific objectives and functions achieved by the present invention, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

[0041] like Figure 1 As shown, a method for generating shoe sole adhesive spraying trajectory based on 3D contours includes the following steps:

[0042] The system acquires point cloud data including the shoe sole. A 3D camera is used to acquire point cloud data of the area where the shoe sole is located, including the shoe sole and environmental data.

[0043] For segmentation processing, the RANSAC random sampling consensus algorithm is used to segment the obtained point cloud data into foreground point cloud and background point cloud. The shoe sole point cloud is the foreground point cloud, thus obtaining the segmented shoe sole point cloud and background point cloud, which facilitates subsequent separate processing of the shoe sole point cloud.

[0044] For orientation processing, the principal component analysis (PCA) method is used to determine the principal orientation of the sole point cloud. The principal orientation of the sole point cloud is then adjusted to be the same as that of the pipeline. If the principal orientation of the sole point cloud is different from that of the pipeline, the sole point cloud is rotated to ensure that the principal orientation of the sole point cloud remains the same as that of the pipeline.

[0045] Resampling process: The shoe sole point cloud that has been oriented is resampled to obtain a resampled secondary shoe sole point cloud;

[0046] Denoising processing: Denoising processing is performed on the secondary shoe sole point cloud;

[0047] The denoised secondary sole point cloud is segmented into three segments based on the direction from the heel to the toe: toe segment, mid-sole segment, and heel segment. The toe segment roughly corresponds to the area from the toe to the forefoot, the heel segment is the semi-circular portion of the heel, and the mid-sole is the portion of the sole excluding the toe and heel segments. The toe and heel segments are roughly semi-circular, while the mid-sole is roughly linear.

[0048] Search for glue spraying points by searching the point clouds of the toe section, the mid-section of the shoe, and the heel section, respectively, to obtain the glue spraying points of the toe section point cloud, the mid-section of the shoe point cloud, and the heel section point cloud.

[0049] By sequentially connecting the glue spraying points at the toe, mid-shoe, and heel, a glue spraying trajectory is formed. Glue is then sprayed along this trajectory to achieve effective bonding of the sole. This provides trajectory data for the robotic arm to spray glue, eliminating the need for manual intervention and improving production efficiency.

[0050] The noise reduction process specifically includes:

[0051] Set the search radius SR and the threshold for the number of neighboring points NP. Traverse each point in the secondary shoe sole point cloud, search for points within the radius SR around the current point, and count the number of searched points TP. If TP is less than NP, then the current point is a noise point and the noise needs to be removed. After traversal, the denoised shoe sole point cloud is obtained.

[0052] The point cloud of the shoe toe segment was searched using the following method:

[0053] like Figure 2As shown, there is a toe and mid-shoe edge line between the toe section and the mid-shoe section. Taking the midpoint of this edge line as the center TC, the center TC divides the toe and mid-shoe edge line into a starting ray TL0 on the left and a terminating ray TLn on the right. The position with the largest jump along the positive direction of the starting ray TL0 is the quasi-adhesive point TP_0. The point with a lower height than quasi-adhesive point TP_0 and a height difference less than TH along the negative direction of the starting ray TL0 is the quasi-adhesive point TP_1. The quasi-adhesive point TP_1 is set as the adhesive point TP0. The starting ray TL0 rotates around the center TC in the opposite direction. The clock hand rotates by a preset angle, forming ray TL1. The search process along the initial ray TL0 is repeated to obtain the glue application point TP1. This process is repeated until the termination ray TLn, located to the right of the center TC, is reached, resulting in all glue application points {TP0, TP1, ..., TPn} in the toe segment point cloud, where n is a natural number. The jump refers to the height difference between the found quasi-glue application point and the center TC. The positive direction is from the center TC along the initial ray TL0, and the negative direction is from outside the initial ray TL0 towards the center TC. The preset angle can be 3 degrees, 5 degrees, or other preset angles. The height difference TH is a constant value. Through this search, an optimal glue application point can be found on a single ray. The search begins from the left side of the center TC and continues until the right side of the center TC, completing the search for the entire toe segment.

[0054] The following method was used to search the point cloud of the shoe heel segment:

[0055] like Figure 3 As shown, there is an edge line between the heel segment and the mid-shoe segment. Taking the midpoint of the edge line as the center BC, the center BC divides the edge line into a starting ray BL0 on the left and a terminating ray BLn on the right. The position with the largest jump along the positive direction of the starting ray BL0 is the quasi-adhesive point BP_0. The point with a lower height than BP_0 and a height difference less than BH along the negative direction of the starting ray BL0 is the quasi-adhesive point BP_1. The quasi-adhesive point BP_1 is set as the adhesive point BP0. The starting ray BL0 is rotated clockwise around the center BC by a preset angle to form ray BL1. The search process along the starting ray BL0 is repeated to obtain the adhesive point BP1. The preset angle is rotated sequentially to the terminating ray BLn on the right side of the center BC to obtain all the adhesive points {BP0, BP1, ..., BPn} of the heel segment point cloud, where n is a natural number. The positive direction of the starting ray BL0 refers to the direction along the starting ray BLO from the center BC. The negative direction of the starting ray BL0 refers to the direction from the outside of the starting ray BL0 toward the center BC. BH is a preset value.

[0056] The search methods for the heel and toe sections are roughly the same: first search from one side of the center, then rotate by a preset angle until you reach the other side, thus completing the overall search. Alternatively, you can start the search from the right side of the center TC or BC, and then rotate to the left.

[0057] The point cloud of the midsection of the shoe was searched using the following method:

[0058] Using the edge line of the toe and mid-shoe sections between the toe and mid-shoe sections as the starting line ML0, and the edge line of the heel and mid-shoe section as the ending line MLn, the midpoint of all point clouds on the starting line ML0 is used as the search starting point MB0. The position with the largest jump to the left along ML0 is the quasi-adhesive spraying point MPL_0. The point with a height lower than MPL_0 and a height difference less than MH along ML0 is used as the quasi-adhesive spraying point MPL_1. The quasi-adhesive spraying point MPL_1 is set to be the adhesive spraying point MPL0, and MH is a preset value.

[0059] Find the position with the largest jump to the right along ML0 as the quasi-adhesion point MPR_0. Search for the point with a height lower than MPR_0 and a height difference less than MH to the left along ML0 as the quasi-adhesion point MPR_1. Set the quasi-adhesion point MPR_1 as the adhesion point MPR0. Translate the starting line ML0 along its orthogonal direction toward the edge of the shoe heel by a preset distance to form line ML1. Repeat the search process along the starting line ML0 to obtain the adhesion point MPL1 on the left and the adhesion point MPR1 on the right. Translate the line by a preset distance to the termination line MLn to obtain all the adhesion point positions {MPL0, MPL1, ..., MPLn} and {MPR0, MPR1, ..., MPRn} of the mid-shoe point cloud.

[0060] For searching the midsection of the shoe, since the midsection is roughly straight, the search line needs to be shifted vertically to achieve a comprehensive search.

[0061] The glue spraying points are accurately located by searching the toe section, mid-section, and heel section of the shoe separately.

[0062] Additionally, a 3D contour-based system for generating adhesive spraying trajectories on shoe soles includes:

[0063] The acquisition unit is used to acquire point cloud data of the shoe sole;

[0064] Segmentation unit, used to segment point cloud data into foreground point cloud and background point cloud;

[0065] Rotation unit, used to rotate the foreground cloud;

[0066] Noise reduction unit, used for noise reduction of shoe sole dot cloud;

[0067] Segmentation units are used to divide the sole point cloud into toe segment point cloud, mid-shoe segment point cloud, and heel segment point cloud;

[0068] The search unit is used to search for glue spray points on the point cloud of the toe section, the point cloud of the mid section, and the point cloud of the heel section.

[0069] It should be noted that the above description is not intended to limit the present invention. Any obvious substitutions without departing from the inventive concept of the present invention are within the protection scope of the present invention.

Claims

1. A method for generating shoe sole adhesive spraying trajectory based on 3D contours, characterized in that, Includes the following steps: Obtain point cloud data including the shoe sole; The point cloud data is segmented into foreground point cloud and background point cloud, with the shoe sole point cloud being the foreground point cloud. Direction processing: Determine the main direction of the sole point cloud and adjust the main direction of the sole point cloud to be the same as the direction of the production line; Resampling process: The shoe sole point cloud that has been oriented is resampled to obtain a resampled secondary shoe sole point cloud; Denoising processing: Denoising processing is performed on the secondary shoe sole point cloud; Segmented processing: The noise-reduced secondary sole point cloud is divided into toe segment point cloud, mid segment point cloud, and heel segment point cloud according to the direction from the heel to the toe of the sole. Search for glue spraying points by searching the point clouds of the toe section, the mid-section of the shoe, and the heel section, respectively, to obtain the glue spraying points of the toe section point cloud, the mid-section of the shoe point cloud, and the heel section point cloud. Connect the glue spraying points at the toe, mid-section, and heel in sequence to form a glue spraying trajectory; The point cloud of the shoe toe segment was searched using the following method: There is a toe and mid-shoe edge line between the toe and mid-shoe sections. The midpoint of the toe and mid-shoe edge line is taken as the center TC. The center TC divides the toe and mid-shoe edge line into a starting ray TL0 on the left and a terminating ray TLn on the right. The position with the largest jump along the positive direction of the starting ray TL0 is the quasi-adhesive point TP_0. The point with a lower height than the quasi-adhesive point TP_0 and a height difference less than TH along the opposite direction of the starting ray TL0 is the quasi-adhesive point TP_1. The quasi-adhesive point TP_1 is set as the adhesive point TP0. The starting ray TL0 is rotated counterclockwise around the center TC by a preset angle to form ray TL1. The search process along the starting ray TL0 is repeated to obtain the adhesive point TP1. The preset angle is rotated sequentially to the terminating ray TLn on the right side of the center TC to obtain all the adhesive point positions {TP0, TP1, ... TPn} of the toe section point cloud. The jump refers to the height difference between the found spray point and the center TC. The positive direction of the starting ray TL0 is the direction along the starting ray TLO from the center TC. The negative direction of the starting ray TL0 is the direction from the outside of the starting ray TL0 toward the center TC.

2. The method for generating shoe sole adhesive spraying trajectory based on 3D contours according to claim 1, characterized in that, The segmentation process employs a random sampling consensus algorithm.

3. The method for generating shoe sole adhesive spraying trajectory based on 3D contours according to claim 1, characterized in that, The direction processing is performed using principal component analysis. If the principal direction of the shoe sole point cloud is different from the pipeline direction, the shoe sole point cloud is rotated to make the principal direction of the shoe sole point cloud the same as the pipeline direction.

4. The method for generating shoe sole adhesive spraying trajectory based on 3D contour according to claim 1, characterized in that, The noise reduction process specifically includes: Set the search radius SR and the threshold for the number of neighboring points NP. Traverse each point in the secondary shoe sole point cloud, search for points within the radius SR around the current point, and count the number of searched points TP. If TP is less than NP, then the current point is a noise point and the noise needs to be removed. After traversal, the denoised shoe sole point cloud is obtained.

5. The method for generating shoe sole adhesive spraying trajectory based on 3D contour according to claim 1, characterized in that, The following method was used to search the point cloud of the shoe heel segment: There is a heel and mid-shoe edge line between the heel and mid-shoe sections. Taking the midpoint of the heel and mid-shoe edge line as the center BC, the center BC divides the heel and mid-shoe edge line into a starting ray BL0 on the left and a terminating ray BLn on the right. The position with the largest jump along the positive direction of the starting ray BL0 is the quasi-adhesive point BP_0. The point with a lower height than BP_0 and a height difference less than BH along the opposite direction of the starting ray BL0 is the quasi-adhesive point BP_1. The quasi-adhesive point BP_1 is set as the adhesive point BP0. The starting ray BL0 is rotated clockwise around the center BC by a preset angle to form ray BL1. The search process along the starting ray BL0 is repeated to obtain the adhesive point BP1. The preset angle is rotated sequentially to the terminating ray BLn on the right side of the center BC to obtain all the adhesive point positions {BP0, BP1, ... BPn} of the heel section point cloud. The positive direction of the initial ray BL0 refers to the direction along the initial ray BLO starting from the center BC, while the negative direction of the initial ray BL0 refers to the direction from the outside of the initial ray BL0 toward the center BC.

6. The method for generating shoe sole adhesive spraying trajectory based on 3D contour according to claim 5, characterized in that, The point cloud of the midsection of the shoe was searched using the following method: Using the edge line of the toe and mid-shoe sections between the toe and mid-shoe sections as the starting line ML0, and the edge line of the heel and mid-shoe sections as the ending line MLn, the midpoint of all point clouds on the starting line ML0 is used as the search starting point MB0. The position with the largest jump to the left along ML0 is the quasi-adhesive point MPL_0. The point with a height lower than MPL_0 and a height difference less than MH is searched to the right along ML0 and is the quasi-adhesive point MPL_1. The quasi-adhesive point MPL_1 is set as the adhesive point MPL0. Find the position with the largest jump to the right along ML0 as the quasi-adhesion point MPR_0. Search for the point with a lower height than MPR_0 and a height difference less than MH to the left along ML0 as the quasi-adhesion point MPR_1. Set the quasi-adhesion point MPR_1 as the adhesion point MPR0. Translate the starting line ML0 along its orthogonal direction towards the edge of the shoe heel by a preset distance to form line ML1. Repeat the search process along the starting line ML0 to obtain the adhesion point MPL1 on the left and the adhesion point MPR1 on the right. Translate the line by a preset distance to the termination line MLn to obtain all the adhesion points {MPL0, MPL1, ..., MPLn} and {MPR0, MPR1, ..., MPRn} of the mid-shoe point cloud.

Images