An insole stamping robot system based on adaptive edge detection and path planning

Abstract

The present application relates to a kind of insole stamping robot system based on adaptive edge detection and path planning.It includes automatic feeding module, mechanical arm module, blanking workbench module;Among them, automatic feeding module is used to realize the automatic feeding of face lining;Mechanical arm module picks up face lining to the blanking workbench of blanking workbench module, and carries out the positioning movement of blanking die of blanking workbench module;Blanking workbench module is used to carry out face lining edge detection positioning, and face lining blanking, while being able to control the action of automatic feeding module and mechanical arm module.The present application is based on BIT trajectory planning and adaptive edge detection algorithm and robot communication and control technology to establish face lining blanking robot system for the convenience of face lining blanking feeding, can support automatic feeding and visual positioning blanking, solve the problem that manual blanking die placement faces great labor intensity, certain safety risk and inaccurate die placement on face lining raw material.

Description

TECHNICAL FIELD

[0001] The present application relates to a shoe insole stamping robot system based on adaptive edge detection and path planning. BACKGROUND

[0002] The existing equipment is old and mainly relies on manual operation. Manual placement of the blanking die faces the problems of high labor intensity, certain safety risks and inaccurate placement of the die on the insole material. Therefore, it is urgent to establish a completely independent and modular insole blanking platform to support automatic feeding and visual positioning blanking, solving the problems of high labor intensity, certain safety risks and inaccurate placement of the die on the insole material. SUMMARY

[0003] The present application aims to solve the problems of high labor intensity, certain safety risks and inaccurate placement of the die on the insole material in manual placement of the blanking die, and therefore provides a shoe insole stamping robot system based on adaptive edge detection and path planning, which is composed of an automatic feeding module, a mechanical arm module and a blanking workbench module.

[0004] To achieve the above-mentioned purpose, the technical solution of the present application is as follows: a shoe insole stamping robot system based on adaptive edge detection and path planning, comprising an automatic feeding module, a mechanical arm module and a blanking workbench module; wherein,

[0005] The automatic feeding module is used to realize automatic feeding and replenishment of the insole.

[0006] The mechanical arm module picks up the insole to the blanking workbench of the blanking workbench module and performs positioning and movement of the blanking die of the blanking workbench module.

[0007] The blanking workbench module is used to perform insole edge detection and insole blanking, and can control the actions of the automatic feeding module and the mechanical arm module.

[0008] In an embodiment of the present application, the insole edge detection performed by the blanking workbench module is realized by an insole edge detection method based on color features and PSO optimized morphological parameters, and the specific implementation is as follows:

[0009] S1, obtain the RGB original image of the insole and perform image preprocessing;

[0010] S2, convert the preprocessed image from the RGB color domain to the HSV color domain;

[0011] S3, after the color domain conversion is completed, define a predetermined blue color range, and create a mask based on this range to mark the area meeting the predetermined blue color range in the original image;

[0012] S4, apply median filtering and dilation morphological operation to the mask, and use PSO algorithm to adaptively optimize parameters of the median filtering and the dilation morphological operation;

[0013] S5, identify edges in the image and perform contour detection: analyze objects in the image, and calculate area of each contour, and screen contours meeting preset standards; for each selected contour, calculate its minimum circumscribed rectangle, and draw the corresponding contour based on the minimum circumscribed rectangle;

[0014] S6, based on the obtained contour, output coordinates and angle of the minimum circumscribed rectangle of the face mask, and complete the entire image processing process.

[0015] In an embodiment of the present application, in step S1, the RGB original image of the face mask is collected by a depth camera arranged above the punching workbench.

[0016] In an embodiment of the present application, in step S2, the preprocessed image is converted from the RGB color domain to the HSV color domain in the following manner:

[0017] Let (r, g, b) be the red, green and blue coordinates of a color, and their values are real numbers between 0 and 1; let max be the maximum of r, g and b; let min be the minimum of r, g and b; then the calculation formula for converting the RGB color domain to the HSV color domain is as follows:

[0018]

[0019]

[0020] v=max

[0021] In the formula, h represents hue, s represents saturation, and v represents brightness value.

[0022] In an embodiment of the present application, in step S4, the objective function of the PSO algorithm is constructed based on two aspects: one is the quantification of noise elimination effect, and the mean square error (MSE) is used to measure the difference between the original image matrix A and the processed image matrix B; the other is to quantify the difference between the areas of the two largest regions in the processed image to evaluate the accuracy of region selection.

[0023] In an embodiment of the present application, the objective function of the PSO algorithm has the form F(A, B) = λ·Fnoise(A, B) + (1-λ)·Farea(R1, R2), where λ is a weighting coefficient used to balance the importance of noise elimination effect and region selection criteria.

[0024] In an embodiment of the present application, in step S5, Gaussian blur and Canny edge detection method are used to identify edges in the image.

[0025] In an embodiment of the present application, the efficient path planning of the mechanical arm of the mechanical arm module in the blanking work is based on the improved FSPS-BIT* achieved by integrating the BIT* algorithm and the FSPS algorithm, and the specific implementation is as follows:

[0026] The set of all points tried for sampling in the BIT*“sampling” function is denoted by X rand , and rejection sampling is used to facilitate the direct calculation of X rand and X sample :

[0027] {X sample ∣x∈X rand ∩X free ,c(X root ,X)+c(X goal ,X)<C best}

[0028]

[0029] Where X sample is the set of sampling points; X root is the set of initial nodes; C best is the set of costs; X free is the result set of allowed states; X goal is the set of desired target states; S is the state space of the path planning problem; c(,) represents the set with the smaller cost; V space is the spatial volume; λ(·) represents the Lebesgue measure of the corresponding set; X f represents the solution that can provide a better cost than the current best solution;

[0030] After obtaining the original path of the BIT* planning, the original path is further simplified by adding the FSPS algorithm, and the specific process of the FSPS algorithm for simplifying the original path is as follows:

[0031] First, the starting point of the original path is used as the initial node of the simplified path; then, the original path is iteratively processed through a while loop and a nested for loop; in the inner loop, the cost function is used to determine whether there is an obstacle between the two nodes, and if there is no obstacle, the farthest subsequent node is connected to the simplified path; if a node cannot be directly connected to the farthest subsequent node, the FSPS algorithm will continue to traverse the original path until all nodes are processed.

[0032] In an embodiment of the present application, the automatic feeding module includes a face lining raw material transport vehicle capable of moving the feeding, and a feeding machine capable of automatically feeding after the face lining is picked up by the mechanical arm module.

[0033] In an embodiment of the present application, the mechanical arm module includes a first mechanical arm for positioning and moving the punching die of the punching workbench module, and a second mechanical arm for picking up the face lining to the punching workbench.

[0034] Compared with the prior art, the present application has the following beneficial effects: the shoe insole punching robot system based on adaptive edge detection and path planning is based on BIT trajectory planning, adaptive edge detection algorithm, robot communication and control technology, and establishes a completely independent and modular face lining punching platform, which can support automatic feeding and visual positioning punching, and solves the problems of high labor intensity, certain safety risk and inaccurate placement of the punching die on the face lining raw material. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 It is a schematic diagram of the shoe insole punching robot system based on adaptive edge detection and path planning.

[0036] Figure 2 It is a flow chart of the face lining die edge detection method based on color features and PSO optimization morphological parameters.

[0037] Figure 3 It is an effect diagram of the face lining die edge detection method based on color features and PSO optimization morphological parameters.

[0038] Figure 4 It is a pseudo code of the original path simplification process of the FSPS algorithm. DETAILED DESCRIPTION

[0039] The technical solutions of the present application will be specifically described below with reference to the drawings.

[0040] The present application provides a shoe insole punching robot system based on adaptive edge detection and path planning, which comprises an automatic feeding module, a mechanical arm module and a punching workbench module.

[0041] The automatic feeding module is used for realizing automatic feeding and replenishment of the face lining.

[0042] The mechanical arm module picks up the face lining to the punching workbench of the punching workbench module, and performs positioning and moving of the punching die of the punching workbench module.

[0043] The punching workbench module is used for performing edge detection of the face lining and punching of the face lining, and can control the actions of the automatic feeding module and the mechanical arm module.

[0044] Specifically, as Figure 1As shown, the present application establishes a face lining blanking robot system with BIT trajectory planning and adaptive edge detection algorithm and robot communication and control technology, and the developed stamping robot system is fast and safe for face lining blanking Figure 1 As shown, it consists of three main modules: automatic feeding module, mechanical arm module and blanking workbench module.

[0045] Firstly, the system picks up the face lining in the material frame by the mechanical arm 2 instead of manual assembly line feeding, and then transports it to the workbench. The D435 depth camera collects the face lining information, and the face lining positioning is carried out. The three-dimensional coordinates of the center of the face lining are transmitted to the mechanical arm 1, and the blanking die is accurately positioned and moved above the face lining. At the same time when the mechanical arm 1 completes the blanking task, the mechanical arm 2 is sent again to pick up the signal, and the face lining processing is completed.

[0046] Based on the D435 depth camera, the face lining information is collected, as shown in the figure. Figure 2 As shown, the present application detects the edge of the face lining die based on color features and PSO optimization morphological parameters, and the specific process is as follows:

[0047] In order to effectively extract the information parameters of the face lining die from the image captured by the depth camera, we propose an image processing algorithm based on color features and PSO optimization morphological parameters. The input of the algorithm is the RGB original image captured by the camera, and the output is the coordinates and angle of the minimum circumscribed rectangle of the face lining die. During the implementation of the algorithm, the first step is image preprocessing, including reading the image and scaling it to reduce the amount of calculation and improve the processing efficiency, while maintaining the original aspect ratio of the image. Next, in order to more effectively extract the color features, the algorithm converts the image from the RGB color domain to the HSV color domain.

[0048] Let (r, g, b) be the red, green and blue coordinates of a color, respectively, and their values are real numbers between 0 and 1; let max be the maximum of r, g, b; let min be the minimum of r, g, b;

[0049]

[0050]

[0051] v = max

[0052] After the color domain conversion is completed, the algorithm defines a specific blue range and creates a mask based on this range to mark the area in the original image that meets the blue range. Subsequently, in order to reduce image noise and enhance key features, the algorithm applies median filtering and dilation morphological operations to the mask.

[0053] In the proposed image processing algorithm, we use the PSO algorithm to adaptively optimize the parameters of filtering and morphological operations. The objective function of the algorithm is constructed based on two main aspects: one is the quantification of noise removal effect, using mean square error (MSE) to measure the difference between the original image matrix A and the processed image matrix B; the second is to quantify the difference between the areas of the two largest regions in the processed image (because the face lining mold region will eventually present as two larger areas), to evaluate the accuracy of region selection. The form of the objective function is F(A, B) = λ·Fnoise(A, B) + (1-λ)·Farea(R1, R2), where λ is the weighting coefficient, used to balance the importance of noise removal effect and region selection criteria. By searching for the optimal solution of this objective function through the PSO algorithm, the best feature parameters for each picture can be obtained.

[0054] After that, the algorithm uses Gaussian blur and Canny edge detection method to identify the edges in the image, laying the foundation for the next contour detection. In the contour detection stage, the algorithm analyzes the objects in the image and calculates the area of each contour, and selects the contours that meet the preset criteria for subsequent processing. For each selected contour, the algorithm calculates its minimum bounding rectangle, and on this basis, the contour is drawn. Finally, the algorithm outputs the coordinates and angles of the minimum bounding rectangle of the face lining mold based on these contours, completing the entire image processing process. The algorithm effect diagram is shown in Figure 3 .

[0055] In order to realize the efficient path planning of the mechanical arm module in the punching work, an improved FSPS-BIT* algorithm based on integrated BIT* algorithm and FSPS algorithm is proposed.

[0056] The original BIT* is a path planning algorithm that balances the features of graph-based search algorithms and sampling-based algorithms. It approximates the environment by maintaining a random geometric graph (RGG), which uses a parameter r to describe the adjacency of the RGG graph. When constructing the RGG graph using multiple batches of sampling, the RGG graph will become denser and more specific to the approximation of the environment. BIT* uses a heuristic based on estimated cost, including the current cost in the RGG graph and the target. BIT* uses an edge queue and a node queue to guide the search. If the performance of the best node in the node queue is better than the best edge in the edge queue, this node will be expanded. Conversely, consider whether the best edge in the edge queue can improve the current search tree, if it can check whether this edge will cause a collision, if there is no conflict, it will be added to the current search tree. If a better solution is found in this iteration, BIT* will use the method in Informed RRT* to improve the planning efficiency of the next batch of sampling by using elliptical sampling.

[0057] An Improved Algorithm FSPS-BIT*: Traditional BIT* path planning algorithms exhibit good performance in narrow environments within high-dimensional spaces, effectively generating collision-free paths. However, they still suffer from the drawback of winding paths. To address this issue, this paper proposes an FSPS-BIT* algorithm to optimize paths. FSPS aims to effectively reduce the number of path nodes in automated path planning while maintaining the overall path shape and accessibility. The algorithm uses the starting point of the original path as the initial node for the simplified path, and then simplifies the path by traversing each node of the original path and attempting to directly connect to the furthest subsequent node. The algorithm pays particular attention to the direct reachability between nodes, using a cost function to determine whether there are obstacles between two nodes. Application results show that FSPS can significantly reduce the number of nodes while maintaining path shape and accessibility, demonstrating significant advantages in computational efficiency and simplification effect. This algorithm is of great significance for improving the efficiency and practicality of autonomous navigation and robot path planning. Future research will explore its applicability and optimization potential in different scenarios, providing an effective path simplification solution for automated path planning.

[0058] The improved FSPS-BIT* algorithm is as follows:

[0059] The improvement to the BIT* algorithm lies in: X is used to represent the set of all points to be sampled in the "sampling" function of BIT*. rand This indicates that rejection sampling is used to facilitate the direct calculation of X. rand and X sample :

[0060] {X sample |x∈X rand ∩X free ,c(X root ,X)+c(X goal ,X)<C best}

[0061]

[0062] Among them, X sample X is the set of sampling points; root C is the initial set of nodes; best For cost set; X free The result set of allowed states; X goal The set of desired target states; Let V be the state space of the path planning problem; c(,) denotes choosing the set with the smaller cost from two sets; space Let X be the spatial volume; λ(·) denotes the Lebesgue measure of the corresponding set; X f This indicates that a solution with a better cost than the current best solution can be provided;

[0063] After obtaining the original path of BIT* planning, the original path is further simplified by adding FSPS algorithm, and the pseudo code of the simplification process is as shown in the following formula Figure 4 The specific process is as follows:

[0064] Firstly, the starting point of the original path is used as the initial node of the simplified path; then, the original path is iteratively processed through while loop and nested for loop; in the inner loop, the cost function is used to judge whether there is an obstacle between two nodes, if there is no obstacle, the farthest subsequent node is connected to the simplified path; if a node cannot be directly connected to the farthest subsequent node, the FSPS algorithm will continue to traverse the original path until all nodes are processed.

[0065] Figure 4 X path represents the original path, i represents the step X simplified simplified path, E represents the path node, Edgeto path() and PathtoEdge() both represent the extraction of the path node.

[0066] This simplification method can effectively reduce the number of redundant path points. This method also reduces the collision detection of long local paths (for example, linear paths from the starting point to the target point). As the path distance increases, the frequency of each collision detection also increases, so avoiding collision detection of long paths can significantly improve the efficiency of the algorithm.

[0067] The above is the preferred embodiment of the present application, any changes made according to the technical solutions of the present application, as long as the generated function does not exceed the scope of the technical solutions of the present application, belongs to the protection scope of the present application.

Claims

1. A shoe insole stamping robot system based on adaptive edge detection and path planning, characterized in that, It includes an automatic feeding module, a robotic arm module, and a blanking worktable module; among which, An automatic feeding module is used to automatically feed and replenish the fabric lining. The robotic arm module picks up the liner and places it onto the blanking worktable of the blanking worktable module, and performs positioning and movement of the blanking die of the blanking worktable module. The blanking workbench module is used for edge detection of the lining and blanking of the lining, and can also control the movements of the automatic feeding module and the robotic arm module. The blanking workbench module performs surface edge detection based on a method that uses color features and PSO-optimized morphological parameters. The specific implementation is as follows: S1. Obtain the original RGB image of the lining and perform image preprocessing; S2. Convert the preprocessed image from the RGB color gamut to the HSV color gamut; S3. After the color gamut conversion is completed, define a predetermined blue range and create a mask based on this range to mark the area in the original image that conforms to the predetermined blue range. S4. Apply median filtering and dilation morphological operations to the mask, and use the PSO algorithm to adaptively optimize the parameters of the median filtering and dilation morphological operations. S5. Identify edges in the image and perform contour detection: Analyze the objects in the image and calculate the area of ​​each contour, and filter out contours that meet the preset criteria; for each selected contour, calculate its minimum bounding rectangle and draw the corresponding contour based on it. S6. Based on the obtained contour, output the coordinates and angles of the minimum bounding rectangle of the surface to complete the entire image processing process. In step S1, the original RGB image of the lining is acquired by a depth camera positioned above the punching table; In step S2, the method for converting the preprocessed image from the RGB color gamut to the HSV color gamut is as follows: Let (r, g, b) be the red, green, and blue coordinates of a color, with values ​​between 0 and 1; let max equal the maximum of r, g, and b; let min equal the minimum of r, g, and b; then the formula for converting the RGB color gamut to the HSV color gamut is as follows: In the formula, h represents hue, s represents saturation, and v represents brightness.

2. The insole stamping robot system based on adaptive edge detection and path planning according to claim 1, characterized in that, In step S4, the objective function of the PSO algorithm is constructed based on two aspects: first, the quantification of noise reduction effect, using mean square error (MSE) to measure the difference between the original image matrix A and the processed image matrix B; second, the quantification of the difference between the two largest regions in the processed image to evaluate the accuracy of region selection.

3. A shoe insole stamping robot system based on adaptive edge detection and path planning according to claim 1 or 2, characterized in that, The objective function of the PSO algorithm is in the form of F(A,B)=λ·Fnoise(A,B)+(1-λ)·Farea(R1,R2), where λ is a weighting coefficient used to balance the importance of noise reduction effect and region selection criteria.

4. The insole stamping robot system based on adaptive edge detection and path planning according to claim 1, characterized in that, In step S5, Gaussian blur and Canny edge detection methods are used to identify edges in the image.

5. The insole stamping robot system based on adaptive edge detection and path planning according to claim 1, characterized in that, The efficient path planning of the robotic arm in the punching process is based on an improved FSPS-BIT* implementation that integrates the BIT* algorithm and the FSPS algorithm, as detailed below: The set of all points that are attempted to be sampled in the "sampling" function of BIT* is represented by X. rand This indicates that rejection sampling is used to facilitate the direct calculation of X. rand and X sample : {X sample ∣x∈X rand ∩X free ,c(X root ,X)+c(X goal ,X)<C best } Among them, X sample X is the set of sampling points; root C is the initial set of nodes; best For cost set; X free The result set of allowed states; X goal The set of desired target states; Let V be the state space of the path planning problem; c(,) denotes choosing the set with the smaller cost from two sets; space Let X be the spatial volume; λ(·) denotes the Lebesgue measure of the corresponding set; X f This indicates that a solution with a better cost than the current best solution can be provided; After obtaining the original path of the BIT* plan, the original path is further simplified by incorporating the FSPS algorithm. The specific process of simplification of the original path by the FSPS algorithm is as follows: First, the starting point of the original path is used as the initial node of the simplified path. Then, the original path is iteratively processed using a while loop and nested for loops. In the inner loop, a cost function is used to determine whether there is an obstacle between two nodes. If there is no obstacle, the farthest subsequent node is connected to the simplified path. If a node cannot be directly connected to the farthest subsequent node, the FSPS algorithm will continue to traverse the original path until all nodes have been processed.

6. The insole stamping robot system based on adaptive edge detection and path planning according to claim 1, characterized in that, The automatic feeding module includes a moving material transport vehicle for fabric lining, and a feeding machine that works in conjunction with the material transport vehicle to automatically feed the fabric lining after it is picked up by the robotic arm module.

7. The insole stamping robot system based on adaptive edge detection and path planning according to claim 1, characterized in that, The robotic arm module includes a first robotic arm for positioning and moving the blanking die of the blanking worktable module, and a second robotic arm for picking up the liner onto the blanking worktable.

Images