Mechanical arm intelligent offset coil grabbing method based on visual system
The intelligent offset grasping method of the robotic arm assisted by the vision system solves the problem of unstable manual positioning accuracy of the stator coil, realizes high-precision and low-damage coil grasping, improves production efficiency and reduces costs.
Patent Information
- Application Number
- CN202510942057.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-09
- Publication Date
- 2025-09-30
AI Technical Summary
In the existing technology, the grasping and positioning operations of the stator coil rely on manual labor, which makes the positioning accuracy susceptible to human factors and difficult to unify. There is a high risk of uncertainty, which may cause the coil to deviate from the grasping range of the robotic arm, causing physical damage and increased production costs.
An intelligent offset grasping method for a robotic arm based on a vision system is adopted. Through image acquisition, preprocessing, feature point extraction, calculation of the distance between feature points and judgment of the gripper width, non-contact perception and high-resolution imaging are achieved. The robotic arm automatically deviates to a safe position to grasp the coil.
It improves positioning accuracy and operation stability, reduces the probability of coil grabbing damage, reduces manual interference, improves production efficiency and reduces production costs.
Smart Images
Figure CN120715892A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of intelligent stator coil grabbing, and in particular to a method for intelligently offsetting coil grabbing using a robotic arm based on a vision system. Background Art
[0002] With the rapid development of industrial automation and intelligent manufacturing, position detection based on vision systems has become a core link in improving production efficiency and ensuring product quality. Therefore, in the production process of stator coils, position detection systems and equipment based on vision systems are often required to assist in completing the production of stator coils.
[0003] The existing positioning operation of grabbing coils mainly relies on manual work. Workers need to use measuring tools to manually adjust the coil position so that it reaches a preset fixed point. The robotic arm then performs the grabbing action according to the established trajectory. However, the positioning accuracy of this manual method is easily affected by human factors and fluctuates greatly. In addition, it is difficult to unify the execution standards among different operators, and there is a high risk of uncertainty. Once the operator makes a mistake, the coil may deviate from the precise grabbing range of the robotic arm, directly causing physical damage to the coil during the grabbing process. This will not only cause production delays, but may also cause the coil to be scrapped, thereby increasing production costs. Summary of the Invention
[0004] The purpose of the present invention is to provide a method for intelligent offset coil grabbing by a robotic arm based on a visual system to solve the problem that the positioning operation of grabbing coils proposed in the above background technology mainly relies on manual completion. The staff needs to manually adjust the position of the coil with the help of measuring tools to make it reach a preset fixed point. Then the robotic arm performs the grabbing action according to the established trajectory. However, this manual positioning accuracy is easily affected by human factors and fluctuates greatly, and the execution standards between different operators are difficult to unify, and there is a high risk of uncertainty. Once the operator makes a mistake, the coil may deviate from the precise grabbing range of the robotic arm, which directly causes the robotic arm to cause physical damage to the coil during the grabbing process, which will not only cause production delays, but may also cause the coil to be scrapped, thereby increasing production costs.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligently offsetting and grasping coils using a robotic arm based on a vision system, comprising image acquisition → image preprocessing → feature point extraction → calculation of the distance between feature points → determination of the distance between feature points and the width of the gripper → grasping the coils. The specific steps of the method for grasping coils are as follows: Step 1: Image acquisition: The robotic arm moves above the transverse tray to take a photo of the non-lead end coil, with the end gripper of the robotic arm facing the lower transverse positioning block. Assign this step a value of 1 in the robotic arm program.
[0006] Step 2: Image preprocessing: Preprocess the captured image, including image denoising and image enhancement. During the image acquisition process, the image will be affected by factors such as light spots and shadows, generating noise. Noise pollution will directly affect the subsequent feature point extraction.
[0007] The specific steps of the coil grabbing method also include: Step 1: Feature point extraction: The extracted feature points include the feature points of the non-lead end coil and the feature points of the transverse positioning block below the coil. Two transverse positioning blocks are placed on each side of a transverse moving tray. Two transverse positioning blocks on a single side fix a coil. Two coils are placed upside down on one tray. The coil can move laterally in the transverse positioning block slot. Therefore, the point where the outer side of the transverse positioning block under the non-lead end contacts the inner side of the coil is set as the transverse positioning block feature point, and the point at the inner corner of the non-lead end coil is set as the non-lead end coil feature point. Since the transverse positioning block is fixed on the tray and does not move, the coil can only move laterally in the positioning block slot. Therefore, before image acquisition, the robotic arm is moved above the transverse positioning block to take a picture of the positioning block and manually preset the transverse positioning block feature points. The characteristic points of the non-lead end coil are extracted based on the inner angle of the coil. The middle part of the coil is a straight line and the end part is a curved part. A point is extracted every 0.5 cm from the characteristic point of the horizontal positioning block toward the non-lead end, and the slope between the characteristic point of the horizontal positioning block and this point is calculated. If the slope is greater than 15°, the point is the characteristic point of the non-lead end coil. If the slope of the two is less than 15%, the point is discarded, and a point is extracted every 0.5 cm from this point. The slope between the point and the characteristic point of the horizontal positioning block is calculated again. The points are taken until the slope between the taken point and the characteristic point of the horizontal positioning block is greater than 15°, then the point is set as the characteristic point of the non-lead end coil.
[0008] Step 2: Calculate the distance between feature points: Since the feature point of the non-lead end coil is taken from the feature point of the lateral positioning block in the direction of the non-lead end straight line, the two feature points are on the same straight line, and the distance between the two feature points can be directly calculated.
[0009] Step 3: Determine the distance between the feature points and the width of the gripper: The width of the gripper at the end of the robotic arm directly affects whether the robotic arm can grab the coil for transportation. If the gripper width is less than the distance between the two feature points, the gripper can drop down to grab the coil for transportation. Calculate the deviation between the distance between the feature points and the gripper width. If the gripper width is greater than the distance between the two feature points, the coil cannot be dropped down. The robotic arm returns to its initial position and moves to the other side of the lead end coil to take a picture. Assign the value of the program to 0 and return to step 2 for image preprocessing. Since the value ≠ 1, the feature points of the lateral positioning block and the lead end coil are extracted according to the method in step 3.
[0010] Step 4: Grab the coil: If the width of the clamp at the non-lead end is less than the distance between the two feature points, the clamp will first move to the top of the feature point of the lateral positioning block, and then offset the feature point toward the non-lead end by the distance between the feature point and the clamp width, and then drop down to grab the coil; if the clamp at the lead end is directly offset by the distance between the lead end coil feature point and the feature point of the lateral positioning block, then drop down to grab the coil and transfer it to the next process.
[0011] Preferably, the image denoising method uses mean filtering to remove noise from the image. Using a filter window of size a*a, the weighted average of the pixel values surrounding the area covered by the filter window is taken, and this average replaces the pixel value of the center point. This method is then applied to each pixel in the coil image in sequence. This method can smooth noise while preserving edge information. Image enhancement uses histogram equalization to transform the grayscale histograms of the high- and low-brightness coil images into a uniform grayscale histogram, increasing the dynamic range of grayscale value differences between pixels and thereby enhancing the overall image contrast.
[0012] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention reduces manual intervention and eliminates human interference. A camera is installed on the robotic arm to take pictures of the support coil. With its non-contact sensing, high-resolution imaging and multi-target synchronous detection capabilities, the robotic arm automatically shifts to a safe position based on deviation feedback to grasp the coil, thereby improving positioning accuracy and operation stability and reducing the probability of coil damage due to manual operation. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 This is a schematic diagram of the overall system structure of the present invention; Figure 2 Schematic diagram of the coil size detection process based on the vision system in the present invention; Figure 3 This is a schematic structural diagram of the finished coil shape of the present invention; Figure 4 This is a structural diagram of the coil placement method on the lateral positioning block in the present invention; Figure 5 A schematic diagram of a structure for extracting feature point positions in the present invention; Figure 6 This is a schematic diagram of the second structure for extracting feature point positions in the present invention. DETAILED DESCRIPTION
[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0015] See also Figures 1-6 , the present invention provides a technical solution: a method for intelligent offset grasping coils by a robotic arm based on a vision system, comprising image acquisition → image preprocessing → feature point extraction → calculation of the distance between feature points → determination of the distance between feature points and the width of the gripper → grasping the coil; Example
[0016] The specific steps of the coil grabbing method are: Step 1: Image acquisition: The robot moves to the top of the traverse tray to take a photo of the coil at the non-lead end. Figure 3 As shown, the end gripper of the robot arm is facing the horizontal positioning block below, and this step is assigned a value of 1 in the robot arm program.
[0017] Step 2: Image preprocessing: The captured image is preprocessed, including image denoising and image enhancement. During the image acquisition process, the image will be affected by factors such as light spots and shadows, generating noise. Noise pollution will directly affect the subsequent feature point extraction. Image denoising uses mean filtering to remove noise in the image. A filter window of size a*a is used to take a weighted average of the pixel values of the surrounding pixels in the area covered by the filter window. The mean value is used to replace the pixel value of the center point. Each pixel in the coil image is processed in turn. This method can smooth noise while retaining edge point information. Image enhancement uses histogram equalization to convert the grayscale histograms of the high-brightness and low-brightness coil images into a uniform grayscale histogram, which increases the dynamic range of grayscale value differences between pixels, thereby enhancing the overall image contrast.
[0018] Step 3: Feature point extraction: The extracted feature points include the feature points of the non-lead end coil and the feature points of the horizontal positioning block below the coil. Two horizontal positioning blocks are placed on both sides of a horizontal moving tray. Two horizontal positioning blocks on one side fix one coil. Two coils are placed upside down on one tray. The coil can move horizontally in the horizontal positioning block slot, such as Figure 4 As shown, the point where the outer side of the lateral positioning block under the non-lead end contacts the inner side of the coil is set as the lateral positioning block feature point, and the point at the inner corner of the non-lead end coil is set as the non-lead end coil feature point, as shown in FIG. Figure 5 and Figure 6As shown. Since the lateral positioning block is fixed on the tray and does not move, the coil can only move laterally in the positioning block slot. Therefore, before image acquisition, the robotic arm is moved above the lateral positioning block to take a photo of the positioning block and manually preset the lateral positioning block feature points. The feature points of the non-lead end coil are extracted based on the inner angle of the coil. The middle part of the coil is a straight line, and the end is a curved part. From the lateral positioning block feature point, a point is extracted every 0.5 cm in the direction of the non-lead end. The slope between the lateral positioning block feature point and this point is calculated. If the slope is greater than 15°, the point is considered the non-lead end coil feature point. If the slope is less than 15%, the point is discarded and a point is extracted every 0.5 cm from this point. The slope between this point and the lateral positioning block feature point is calculated again. This process continues until the slope between the extracted point and the lateral positioning block feature point is greater than 15°. The extracted point is then set as the non-lead end coil feature point.
[0019] Step 4: Calculate the distance between feature points: Since the feature point of the non-lead end coil is taken from the feature point of the lateral positioning block in the direction of the non-lead end straight line, the two feature points are on the same straight line, and the distance between the two feature points can be directly calculated.
[0020] Step 5: Determine the distance between the feature points and the width of the gripper: The width of the gripper at the end of the robotic arm directly affects whether the robotic arm can grab the coil for transportation. If the gripper width is less than the distance between the two feature points, the gripper can drop down to grab the coil for transportation. Calculate the deviation between the distance between the feature points and the gripper width. If the gripper width is greater than the distance between the two feature points, the coil cannot be dropped down. The robotic arm returns to its initial position and moves to the other side of the lead end coil to take a picture. Assign the value of the program to 0 and return to step 2 for image preprocessing. Since the value ≠ 1, the feature points of the lateral positioning block and the lead end coil are extracted according to the method in step 3.
[0021] Step 6: Grab the coil: If the width of the clamp at the non-lead end is less than the distance between the two feature points, the clamp will first move to the top of the feature point of the lateral positioning block, and then offset the feature point toward the non-lead end by the distance between the feature point and the clamp width, and then drop down to grab the coil; if the clamp at the lead end is directly offset by the distance between the lead end coil feature point and the feature point of the lateral positioning block, then drop down to grab the coil and transfer it to the next process.
[0022] Working principle: First, the staff installs a camera on the six-axis robot arm to identify the position of the coil, and judges the size of the robot arm's gripper by identifying the distance between the feature points, and then grabs the coil on the horizontal positioning block. Then, the robot arm moves to the top of the horizontal moving tray to take a photo of the non-lead end coil. The coil is as follows: Figure 3As shown, the end gripper of the robotic arm is facing the horizontal positioning block below, and this step is assigned a value of 1 in the robotic arm program; the captured image is preprocessed, including image denoising and image enhancement. During the image acquisition process, the image will be affected by, for example, light spots and shadows to generate noise, and the noise pollution will directly affect the subsequent feature point extraction. Image denoising uses mean filtering to remove noise in the image. A filter window of a*a size is used to weighted average the pixel values of the surrounding pixels in the area covered by the filter window, and the mean is used to replace the pixel value of the center point. Each pixel in the coil image is processed in turn. This method can smooth the noise while retaining the edge point information. Image enhancement uses histogram equalization processing to transform the grayscale histogram of the high-brightness and low-brightness coil image into a grayscale uniform histogram, so that the dynamic range of the grayscale value difference between each pixel becomes larger, thereby enhancing the overall contrast of the image; the extracted feature points include the feature points of the non-lead end coil and the feature points of the horizontal positioning block under the coil. Two horizontal positioning blocks are placed on both sides of a horizontal moving tray. Two horizontal positioning blocks on one side fix a coil. A tray can hold two coils placed upside down. The coil can move horizontally in the horizontal positioning block slot, such as Figure 4 As shown, the point where the outer side of the lateral positioning block under the non-lead end contacts the inner side of the coil is set as the lateral positioning block feature point, and the point at the inner corner of the non-lead end coil is set as the non-lead end coil feature point, as shown in FIG. Figure 5 and Figure 6As shown in the figure, since the lateral positioning block is fixed on the pallet and does not move, the coil can only move laterally in the positioning block slot. Therefore, before image acquisition, the robotic arm is moved above the lateral positioning block to take a picture of the positioning block and manually preset the feature points of the lateral positioning block. The characteristic points of the non-lead end coil are extracted according to the inner angle of the coil. The middle part of the coil is a straight line and the end part is a curved part. A point is extracted every 0.5 cm from the characteristic point of the lateral positioning block in the direction of the non-lead end, and the slope between the characteristic point of the lateral positioning block and this point is calculated. If the slope is greater than 15°, the point is the characteristic point of the non-lead end coil. If the slope of the two is less than 15%, the point is discarded, and a point is extracted every 0.5 cm from the point, and the slope between the point and the characteristic point of the lateral positioning block is calculated again. The points are taken until the slope between the taken point and the characteristic point of the lateral positioning block is greater than 15°, then the point is stopped and set as the characteristic point of the non-lead end coil. Since the characteristic point of the non-lead end coil is a point taken according to the characteristic point of the lateral positioning block in the direction of the non-lead end straight line, the two are on the same straight line, and the distance between the two characteristic points can be directly calculated. The width of the gripper at the end of the robotic arm directly affects the robotic arm's ability to move. No, grab the coil for transfer. If the width of the clamp is less than the distance between the two feature points, the clamp can drop to grab the coil for transfer. Calculate the deviation between the distance between the feature points and the width of the clamp. If the width of the clamp is greater than the distance between the two feature points, the coil cannot be dropped. The robot returns to the initial position and moves to the other side of the lead end coil again to take a picture. The program is assigned a value of 0 and returns to step 2 for image preprocessing. Since the value ≠ 1, the feature points of the lateral positioning block and the lead end coil are extracted according to the method in step 3. At the non-lead end, if the width of the clamp is less than the distance between the two feature points, the clamp first moves to the top of the feature point of the lateral positioning block, and then offsets the feature point toward the non-lead end by the size of the deviation between the feature point and the width of the clamp, and drops to grab the coil. If the clamp directly offsets the distance between the lead end coil feature point and the lateral positioning block feature point at the lead end, the coil is dropped and transferred to the next process.
[0023] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.
Claims
1. A method for intelligently offsetting and grasping a coil by a robotic arm based on a vision system, comprising: Image acquisition → image preprocessing → feature point extraction → calculation of the distance between feature points → determination of the distance between feature points and the width of the gripper → coil grabbing. The method for grabbing the coil comprises the following steps: Step 1: Image acquisition: The robotic arm moves above the traverse tray to take a photo of the coil at the non-lead end, with the end gripper of the robotic arm facing the lower lateral positioning block. Assign this step a value of 1 in the robotic arm program. Step 2: Image preprocessing: Preprocess the captured image, including image denoising and image enhancement. During the image acquisition process, the image will be affected by factors such as light spots and shadows, generating noise. Noise pollution will directly affect the subsequent feature point extraction.
2. The method for intelligently offsetting and grasping coils by a robotic arm based on a vision system according to claim 1, characterized in that: The specific steps of the coil grabbing method also include: Step 1: Feature point extraction: The extracted feature points include the feature points of the non-lead end coil and the feature points of the transverse positioning block below the coil. Two transverse positioning blocks are placed on each side of a transverse moving tray. Two transverse positioning blocks on a single side fix a coil. Two coils are placed upside down on one tray. The coil can move laterally in the transverse positioning block slot. Therefore, the point where the outer side of the transverse positioning block under the non-lead end contacts the inner side of the coil is set as the transverse positioning block feature point, and the point at the inner corner of the non-lead end coil is set as the non-lead end coil feature point. Since the transverse positioning block is fixed on the tray and does not move, the coil can only move laterally in the positioning block slot. Therefore, before image acquisition, the robotic arm is moved above the transverse positioning block to take a picture of the positioning block and manually preset the transverse positioning block feature points. The characteristic points of the non-lead end coil are extracted based on the inner angle of the coil. The middle part of the coil is a straight line and the end part is a curved part. A point is extracted every 0.5 cm from the characteristic point of the horizontal positioning block toward the non-lead end, and the slope between the characteristic point of the horizontal positioning block and this point is calculated. If the slope is greater than 15°, the point is the characteristic point of the non-lead end coil. If the slope of the two is less than 15%, the point is discarded, and a point is extracted every 0.5 cm from this point. The slope between the point and the characteristic point of the horizontal positioning block is calculated again. The points are extracted in this way until the slope between the extracted point and the characteristic point of the horizontal positioning block is greater than 15°, then the point is set as the characteristic point of the non-lead end coil. Step 2: Calculate the distance between feature points: Since the non-lead end coil feature point is taken from the lateral positioning block feature point in the non-lead end straight line direction, the two feature points are on the same straight line, and the distance between the two feature points can be directly calculated; Step 3: Determine the distance between the feature points and the width of the gripper: The width of the gripper at the end of the robotic arm directly affects whether the robotic arm can grab the coil for transportation. If the gripper width is less than the distance between the two feature points, the gripper can drop down to grab the coil for transportation. Calculate the deviation between the distance between the feature points and the gripper width. If the gripper width is greater than the distance between the two feature points, the gripper cannot drop down to grab the coil. The robotic arm returns to its initial position and moves to the other side of the lead end coil to take a picture. The program is assigned a value of 0 and returns to step 2 for image preprocessing. Since the value ≠ 1, the feature points of the lateral positioning block and the lead end coil are extracted according to the method in step 3. Step 4: Grab the coil: If the width of the clamp at the non-lead end is less than the distance between the two feature points, the clamp will first move to the top of the feature point of the lateral positioning block, and then offset the feature point toward the non-lead end by the distance between the feature point and the clamp width, and then drop down to grab the coil; if the clamp at the lead end is directly offset by the distance between the lead end coil feature point and the feature point of the lateral positioning block, then drop down to grab the coil and transfer it to the next process.
3. The method for intelligently offsetting and grasping coils by a robotic arm based on a vision system according to claim 1, characterized in that: Image denoising uses mean filtering to remove noise from the image. Using a filter window of size a*a, the weighted average of the pixel values surrounding the area covered by the filter window is taken, replacing the pixel value of the center point with this mean. This process is repeated for each pixel in the coil image. This method smooths noise while preserving edge information. Image enhancement uses histogram equalization to transform the grayscale histograms of the high- and low-brightness coil images into a uniform grayscale histogram. This increases the dynamic range of grayscale value differences between pixels, thereby enhancing overall image contrast.