Mechanical arm position detection device and detection method

Abstract

The invention provides a mechanical arm position detection device and method, and the device comprises a camera element which is disposed at a part, close to an end effector, of a mechanical arm, and is used for collecting a detection image in real time when the end effector picks and places a wafer; and the detection assembly is used for identifying the position of the end effector in the detection image so as to judge whether the position of the end effector is abnormal or not. Compared with a traditional visual method, according to the scheme, the abnormal position condition of the end effector of the mechanical arm can be monitored more accurately, mechanical scratches of the end effector to wafers are reduced, and the problems of yield loss, wafer breakage or wafer scrapping and the like are solved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor manufacturing technology, and in particular to a robot arm position detection device and a detection method. Background Art

[0002] During the wafer production process, the robotic arm on the production equipment reaches into the wafer box to grab the wafer. The parts of the robotic arm will wear due to long-term use, causing the arm to scratch the wafer due to abnormal height or tilt when grabbing or placing the wafer. This mechanical scratch can cause yield loss at the very least, and may cause the wafer to break or be scrapped at the worst.

[0003] In the prior art, in order to prevent the robotic arm from easily scratching the wafer, the visual method is usually used to calibrate the height of the robotic arm. However, the machine itself does not have a device that can be used to monitor abnormal height or tilt of the robotic arm, resulting in wafer scratches still occurring after inspection. In addition, since the visual method cannot be quantified, the inspection standards of each person cannot be completely consistent, which can easily lead to errors. Therefore, the visual method cannot solve the above problems well.

[0004] In view of this, it is necessary to propose a robot arm position detection device and detection method to solve the above problems. Summary of the invention

[0005] The object of the present invention is to provide a robot arm position detection device and detection method, so as to improve the problem that the existing visual method cannot accurately monitor the abnormal position of the robot arm.

[0006] The present invention provides a robot arm position detection device, comprising: An imaging element is provided on the part of the robot arm close to the end effector and is used to collect detection images in real time during the process of the end effector taking and placing the wafer; The detection component is used to identify the position of the end effector in the detection image to determine whether the position of the end effector is abnormal.

[0007] The beneficial effect of the robot arm position detection device provided by the present invention is that by setting the camera element on the end effector of the robot arm, during the process of the end effector taking and placing the wafer, the camera element can collect the detection image containing the end effector in real time, and the detection component performs image processing and analysis on the collected detection image to identify the position of the end effector in the detection image to determine whether the position of the end effector is abnormal. Compared with the traditional visual method, the detection device can provide more accurate detection results, reduce misjudgment caused by human error, and effectively reduce the risk of wafer scratches.

[0008] In a possible embodiment, the detection component includes: An image recognition module, connected to the imaging element and used to identify an actuator recognition area from the detection image, wherein the actuator recognition area is a minimum rectangular area in the detection image that includes the end effector; The position determination module is connected to the image recognition module and is used to determine whether the position of the end effector is abnormal according to the width and length of the effector recognition area.

[0009] Its beneficial effects are: the image recognition module performs image processing on the detection image to identify the actuator recognition area, the position judgment module identifies and determines the width and length of the acquired actuator recognition area, and by analyzing the width and length of the actuator recognition area, it is determined whether the position of the end actuator is abnormal.

[0010] In a possible embodiment, the position determination module is further used to determine whether the actuator identification area satisfies all of the following determination conditions: Determining whether the width of the actuator recognition area is within a set width range; Determining whether the length of the actuator recognition area is within a set length range; If so, it is determined that the position of the end effector is normal; if not, it is determined that the position of the end effector is abnormal.

[0011] Its beneficial effect is that by comparing the width of the actuator recognition area with the width of the actuator recognition area when it is in the standard position, and by comparing the length of the actuator recognition area with the length of the actuator recognition area when it is in the standard position, it is possible to determine whether the end actuator is in the standard position.

[0012] In a possible embodiment, the detection device further includes: A feedback adjustment module is connected to the position judgment module and the robotic arm and is used to control the robotic arm to adjust the position of the end effector until the width of the actuator recognition area is within the set width range and the length of the actuator recognition area is within the set length range when the position judgment module judges that the position of the end effector is abnormal.

[0013] Its beneficial effect is that when the position judgment module determines that the position of the end effector is abnormal, the feedback adjustment module controls the robotic arm to perform rotation correction to adjust the position of the end effector until the width of the actuator recognition area is within the set width range and the length of the actuator recognition area is within the set length range, thereby adjusting the end effector position to normal.

[0014] In a possible embodiment, the image recognition module is further used to identify a wafer recognition area from the detection image, where the wafer recognition area is the smallest rectangular area in the detection image that contains the wafer; When taking the wafer, the judgment condition of the actuator recognition area also includes: Determine whether a difference between a height distance between a left edge point of an upper edge of the actuator identification region and a left edge point of a lower edge of the wafer identification region and a height distance between a center point of an upper edge of the actuator identification region and a center point of a lower edge of the wafer identification region is within a first error range; Determine whether the difference between the height distance between the right edge point of the upper edge of the actuator identification area and the right edge point of the lower edge of the wafer identification area and the height distance between the center point of the upper edge of the actuator identification area and the center point of the lower edge of the wafer identification area is within the first error range.

[0015] Its beneficial effect is that for the detection image collected during the process of picking up the wafer, by comparing the height distance between the two edge points of the actuator identification area and the two edge points of the wafer identification area with the height distance between the center point of the actuator identification area and the center point of the wafer identification area, it is possible to determine whether the difference in height distance between different position points of the actuator identification area and the corresponding position points of the wafer identification area is within the first error range, and then determine whether the position of the end actuator is abnormal.

[0016] In a possible embodiment, the image recognition module is further used to identify a wafer recognition area from the detection image, where the wafer recognition area is the smallest rectangular area in the detection image that contains the wafer; When taking the wafer, the judgment condition of the actuator recognition area also includes: Determine whether the difference between the sum of the height of the left edge point of the upper edge of the actuator identification area and the left edge point of the lower edge of the wafer identification area and the sum of the height of the center point of the upper edge of the actuator identification area and the center point of the lower edge of the wafer identification area is within a second error range; Determine whether the difference between the sum of the height of the right edge point of the upper edge of the actuator identification area and the right edge point of the lower edge of the wafer identification area and the sum of the height of the center point of the upper edge of the actuator identification area and the center point of the lower edge of the wafer identification area is within the second error range.

[0017] Its beneficial effect is that for the detection image collected during the process of picking up the wafer, the sum of the height of the left edge point of the upper edge of the actuator identification area and the left edge point of the lower edge of the wafer identification area, the sum of the height of the right edge point of the upper edge of the actuator identification area and the right edge point of the lower edge of the wafer identification area, and the sum of the height of the center point of the upper edge of the actuator identification area and the center point of the lower edge of the wafer identification area are compared respectively, so that the positional relationship between different position points of the actuator identification area and the corresponding position points of the wafer identification area can be determined, and then whether the position of the end actuator is abnormal can be determined.

[0018] In a possible embodiment, the image recognition module is further used to identify a wafer recognition area from the detection image, where the wafer recognition area is the smallest rectangular area in the detection image that contains the wafer; When placing the wafer, the judgment condition of the actuator identification area also includes: It is determined whether a difference between a height of a center point of an upper edge of the actuator recognition area and a height of a center point of a lower edge of the wafer recognition area is within a third error range.

[0019] Its beneficial effect is that for the detection image collected during the placement of the wafer, by comparing the height of the center point of the upper edge of the actuator identification area with the height of the center point of the lower edge of the wafer identification area, it is determined whether the center point of the upper edge of the actuator identification area coincides with the center point of the lower edge of the wafer identification area, and then it is determined whether the position of the end actuator is abnormal.

[0020] In a possible embodiment, the detection device further includes an alarm connected to the detection component, and when the detection component determines that the position of the end effector is abnormal, the alarm is triggered to sound an alarm.

[0021] Its beneficial effect is that when the detection component determines that the position of the end effector is abnormal, the alarm is triggered to sound an alarm, thereby promptly reminding the staff to monitor the current process of the robotic arm transporting the wafer and take timely measures.

[0022] In a possible embodiment, the detection component is a pre-trained neural network model.

[0023] The present invention also provides a method for detecting the position of a robot arm, comprising: Using the robot arm position detection device as in any of the above embodiments, the camera element is arranged on the end effector of the robot arm; During the detection process, the camera element collects the detection image of the end effector in real time and sends it to the detection component. The detection component processes each of the detection images, identifies the position of the end effector in the detection image, so as to determine whether the position of the end effector is abnormal. If the position of the end effector is determined to be abnormal, the position of the end effector is adjusted until the detection component determines that the position of the end effector is normal. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a three-dimensional diagram of the robot arm position detection device of the present invention.

[0025] Figure 2 It is a schematic diagram of the end arm segment, the end effector and the camera element in the robot arm position detection device of the present invention.

[0026] Figure 3 It is a logic block diagram of the robot arm position detection device of the present invention.

[0027] Figure 4 This is a schematic diagram of the end effector of the robot arm position detection device of the present invention entering a wafer box to grab a wafer.

[0028] Figure 5 This is a state diagram of the robot arm position detection device of the present invention when the end effector is in a normal state and grabs a wafer.

[0029] Figure 6 This is a state diagram of grabbing a wafer when the end effector of the robot arm position detection device of the present invention is in an abnormal state.

[0030] Figure 7 This is a schematic diagram of the end effector of the robot arm position detection device of the present invention entering a wafer box to place a wafer.

[0031] Figure 8 This is a state diagram of placing a wafer when the end effector of the robot arm position detection device of the present invention is in an abnormal state.

[0032] Explanation of the accompanying drawings: 110, camera element; 120, robotic arm; 121, end effector; 122, flip drive mechanism; 123, arm segment; 130, detection component; 131, image recognition module; 132, position judgment module; 133, feedback adjustment module; 140, alarm; 200, wafer; 300, detection image; 310, actuator recognition area; 320, wafer recognition area. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0034] During long-term use, the parts of the robot arm will gradually wear out due to frequent mechanical movements. This wear may cause the robot arm's accuracy to decrease, for example, the robot arm's extension, rotation and grasping movements become less precise. Because the height or angle of the robot arm is not precisely controlled and corrected, when it reaches into the wafer box to grasp or place the wafer, the robot arm may contact and scratch the wafer surface.

[0035] In view of the problems existing in the prior art, an embodiment of the present invention provides a robot arm position detection device, see Figure 1 , Figure 2 as well as Figure 3 The detection device includes an imaging element 110 and a detection component 130. The imaging element 110 is disposed on the robot arm 120 near the end effector 121 and is used to collect the detection image 300 in real time during the process of the end effector 121 taking and placing the wafer 200. For example, the imaging element 110 is a micro camera. The detection component 130 is used to identify the position of the end effector 121 in the detection image 300 to determine whether the position of the end effector 121 is abnormal.

[0036] See also Figure 1 The robot arm 120 includes a plurality of arm segments 123, and two adjacent arm segments 123 are rotatably connected. The portion of the robot arm 120 that takes the wafer 200 is the end arm segment 123. Figure 1 The uppermost arm segment 123 is shown in FIG. Figure 2 The imaging element 110 is disposed on one side surface of the end arm segment 123 close to the end effector 121. For example, the imaging element 110 is embedded in one side surface of the end arm segment 123 close to the end effector 121. A flipping driving mechanism 122 is also disposed in the end arm segment 123. The flipping driving mechanism 122 is connected to the end effector 121 and is used to drive the end effector 121 to rotate. The flipping driving mechanism 122 is a motor, a rotary cylinder, etc.

[0037] In this embodiment, the end effector 121 of the robot arm 120 moves with the imaging element 110 thereon, so that the detection image 300 of the end effector 121 in the process of taking and placing the wafer 200 can be captured in real time. During the process of the robot arm 120 taking and placing the wafer 200, the imaging element 110 will continuously collect the detection image 300. The detection component 130 identifies and analyzes the position of the end effector 121 in the detection image 300 collected by using the image processing algorithm to determine whether the position of the end effector 121 of the robot arm 120 is abnormal, for example, whether the end effector 121 has a height abnormality or tilt. The detection component 130 performs millisecond-level analysis on each frame of the image, and once the position of the end effector 121 is detected to be abnormal, an alarm or automatic adjustment program can be immediately triggered. Compared with the traditional visual method, this solution can provide more accurate monitoring of the abnormal position of the end effector 121 of the robot arm 120, reduce the mechanical scratches of the end effector 121 on the wafer 200, and reduce the problems of yield loss, wafer 200 breakage or wafer 200 scrapping.

[0038] In one embodiment, see Figure 3 , Figure 6 as well as Figure 8 , the detection component 130 includes: An image recognition module 131 is connected to the imaging element 110 and is used to recognize an actuator recognition area 310 from the detection image 300 , where the actuator recognition area 310 is a minimum rectangular area in the detection image 300 that includes the end effector 121 ; The position determination module 132 is connected to the image recognition module 131 and is used to determine whether the position of the end effector 121 is abnormal according to the width and length of the effector recognition area 310 .

[0039] In this embodiment, the image recognition module 131 recognizes the minimum rectangular area of ​​the end effector 121 and the minimum rectangular area of ​​the wafer 200 from the detection image 300 based on an image processing algorithm, etc., so as to quickly and accurately recognize the regional positions of the end effector 121 and the wafer 200 in the detection image 300. The position determination module 132 can determine whether the end effector 121 is tilted by analyzing the width and length of the actuator identification area 310, and further determine whether the position of the end effector 121 is abnormal.

[0040] See also Figure 5 , Figure 6 as well as Figure 8, an XY axis coordinate system is established in the detection image 300, the center point H1 of the lower edge of the wafer identification area 320, the center point H2 of the upper edge of the actuator identification area 310, the left edge point H3 of the lower edge of the wafer identification area 320, the left edge point H4 of the upper edge of the actuator identification area 310 with coordinates (x4, y4), the right edge point H5 of the lower edge of the wafer identification area 320, and the right edge point H6 of the upper edge of the actuator identification area 310 with coordinates (x6, y6).

[0041] In one embodiment, see Figure 3 , Figure 6 as well as Figure 8 The position determination module 132 is also used to determine whether the actuator identification area 310 satisfies all of the following determination conditions: Determine whether the width y2 of the actuator recognition area 310 is within a set width range; Determine whether the length of the actuator identification area 310 is within a set length range, wherein the length of the actuator identification area 310 can be obtained by calculating the difference between the horizontal coordinate x6 of the right edge point H6 of the upper edge of the actuator identification area 310 and the horizontal coordinate x4 of the left edge point H4 of the upper edge of the actuator identification area 310; If so, it is determined that the position of the end effector 121 is normal; if not, it is determined that the position of the end effector 121 is abnormal.

[0042] In this embodiment, it is measured in advance through experiments that when the end effector 121 is in the standard position range, that is, the end effector 121 is not tilted in the vertical direction or is slightly tilted within a reasonable deviation, and is not tilted in the horizontal direction or is slightly tilted within a reasonable deviation, a plurality of detection images 300 are collected by the camera element 110, and a set width range and a set length range are obtained according to the width y2 and length of the actuator recognition area 310 in the plurality of detection images 300. By comparing the width y2 of the actuator recognition area 310 with the set width range of the actuator recognition area 310 in the standard position, and comparing the length of the actuator recognition area 310 with the set length range of the actuator recognition area 310 in the standard position, it is determined whether the end effector 121 is in the standard position state, and further whether the position of the end effector 121 is abnormal, such as Figure 5 and Figure 8 The figure shows the state when the end effector 121 is in a normal position. Figure 6 The figure shows the state when the position of the end effector 121 is abnormal.

[0043] In a specific embodiment, the position judgment module 132 is also used to calculate the deviation under each judgment condition when judging the actuator identification area 310. For example, the deviation includes the width deviation of the width y2 of the actuator identification area 310 exceeding the set width range and the length deviation of the length of the actuator identification area 310 exceeding the set length range.

[0044] In one embodiment, see Figure 3 , the detection device also includes: The feedback adjustment module 133 is connected to the position judgment module 132 and the robot arm 120 and is used to control the robot arm 120 to adjust the position of the end effector 121 according to each deviation amount when the position judgment module 132 determines that the position of the end effector 121 is abnormal until the width y2 of the actuator recognition area 310 is within the set width range and the length of the actuator recognition area 310 is within the set length range.

[0045] Specifically, when adjusting the position of the end effector 121 , the feedback adjustment module 133 controls the flipping drive mechanism 122 of the robot arm 120 to adjust the position of the end effector 121 to a normal position state, that is, a non-tilted state.

[0046] In this embodiment, once an abnormal position of the end effector 121 is detected, the feedback adjustment module 133 will immediately start the adjustment mechanism and control the robot arm 120 to perform rotational correction to adjust the position of the end effector 121 until the width y2 of the actuator identification area 310 is within the set width range and the length of the actuator identification area 310 is within the set length range, and the end effector 121 is adjusted to the standard position, thereby timely adjusting the position of the end effector 121 to a normal state, avoiding problems such as scratches on the wafer 200 caused by position deviation.

[0047] In one embodiment, see Figure 3 , Figure 4 , Figure 5 as well as Figure 6 , the image recognition module 131 is also used to identify a wafer recognition area 320 from the detection image 300 , where the wafer recognition area 320 is the smallest rectangular area in the detection image 300 that contains the wafer 200; When taking the wafer 200, the judgment conditions of the actuator recognition area 310 also include: Determine whether the difference between the height distance between the left edge point H4 of the upper edge of the actuator identification area 310 and the left edge point H3 of the lower edge of the wafer identification area 320 and the height distance between the center point H2 of the upper edge of the actuator identification area 310 and the center point H1 of the lower edge of the wafer identification area 320 is within a first error range; Determine whether the difference between the height distance between the right edge point H6 of the upper edge of the actuator identification area 310 and the right edge point H5 of the lower edge of the wafer identification area 320 and the height distance between the center point H2 of the upper edge of the actuator identification area 310 and the center point H1 of the lower edge of the wafer identification area 320 is within a first error range.

[0048] Specifically, the first error range is between -k and k, and the specific value of k is set according to experience and process requirements.

[0049] In this embodiment, for the detection image 300 collected during the process of taking the wafer 200, the positional relationship between the actuator identification area 310 and the wafer identification area 320 is combined to make a judgment, by comparing the height distance between the left edge point H4 of the upper edge of the actuator identification area 310 and the left edge point H3 of the lower edge of the wafer identification area 320 with the height distance between the center point H2 of the upper edge of the actuator identification area 310 and the center point H1 of the lower edge of the wafer identification area 320, so as to judge whether the distance between the left side of the actuator identification area 310 and the left side of the wafer identification area 320 is equal to the center point H2 of the actuator identification area 310. and the center point of the wafer identification area 320 is equal or within a very small error range; by comparing the height distance between the right edge point H6 of the upper edge of the actuator identification area 310 and the right edge point H5 of the lower edge of the wafer identification area 320 with the height distance between the center point H2 of the upper edge of the actuator identification area 310 and the center point H1 of the lower edge of the wafer identification area 320, it is determined whether the distance between the right side of the actuator identification area 310 and the right side of the wafer identification area 320 is equal to the distance between the center point of the actuator identification area 310 and the center point of the wafer identification area 320 or within a very small error range. In short, the purpose of the above judgment can be understood as judging whether the distances between the left, middle and right positions of the actuator identification area 310 and the wafer identification area 320 are equal or within a very small error range. If so, it is determined that the position of the end effector 121 is abnormal.

[0050] In another embodiment, see Figure 3 , Figure 4 , Figure 5 as well as Figure 6 , the image recognition module 131 is also used to identify a wafer recognition area 320 from the detection image 300 , where the wafer recognition area 320 is the smallest rectangular area in the detection image 300 that contains the wafer 200; When taking the wafer 200, the judgment conditions of the actuator recognition area 310 also include: Determine whether the difference between the sum of the height of the left edge point H4 of the upper edge of the actuator identification area 310 and the left edge point H3 of the lower edge of the wafer identification area 320 and the sum of the height of the center point H2 of the upper edge of the actuator identification area 310 and the center point H1 of the lower edge of the wafer identification area 320 is within a second error range; It is determined whether the difference between the sum of the height of the right edge point H6 of the upper edge of the actuator identification area 310 and the right edge point H5 of the lower edge of the wafer identification area 320 and the sum of the height of the center point H2 of the upper edge of the actuator identification area 310 and the center point H1 of the lower edge of the wafer identification area 320 is within a second error range.

[0051] Specifically, the second error range is between -m and m, and the specific value of m is set based on experience and process requirements.

[0052] In this embodiment, for the detection image 300 collected during the process of taking the wafer 200, the positional relationship between the actuator identification area 310 and the wafer identification area 320 is combined to make a judgment, by comparing the sum of the height of the left edge point H4 of the upper edge of the actuator identification area 310 and the height of the left edge point H3 of the lower edge of the wafer identification area 320 with the sum of the height of the center point H2 of the upper edge of the actuator identification area 310 and the height of the center point H1 of the lower edge of the wafer identification area 320, so as to judge whether the positional relationship between the left side of the actuator identification area 310 and the left side of the wafer identification area 320 is consistent with the positional relationship between the actuator identification area 310 and the left side of the wafer identification area 320. Whether the positional relationship between the center point of the actuator identification area 310 and the center point of the wafer identification area 320 matches; by comparing the sum of the height of the right edge point H6 of the upper edge of the actuator identification area 310 and the height of the right edge point H5 of the lower edge of the wafer identification area 320 with the sum of the height of the center point H2 of the upper edge of the actuator identification area 310 and the height of the center point H1 of the lower edge of the wafer identification area 320, to determine whether the positional relationship between the spacing between the right side of the actuator identification area 310 and the right side of the wafer identification area 320 matches the positional relationship between the center point of the actuator identification area 310 and the center point of the wafer identification area 320. In short, the purpose of the above judgment can be understood as judging whether the positional relationship between the spacings at the left, middle and right positions between the actuator identification area 310 and the wafer identification area 320 matches. If not, it is determined that the position of the end effector 121 is abnormal.

[0053] In one embodiment, see Figure 3 , Figure 7 as well as Figure 8 , the image recognition module 131 is also used to identify a wafer recognition area 320 from the detection image 300 , where the wafer recognition area 320 is the smallest rectangular area in the detection image 300 that contains the wafer 200; When placing the wafer 200, the judgment conditions of the actuator recognition area 310 also include: It is determined that the difference between the height of the center point H2 of the upper edge of the actuator recognition area 310 and the height of the center point H1 of the lower edge of the wafer recognition area 320 is within a third error range.

[0054] Specifically, the third error range is between -n and n, and the specific value of n is set according to experience and process requirements.

[0055] In this embodiment, for the detection image 300 collected during the placement of the wafer 200, a judgment is made in combination with the positional relationship between the actuator identification area 310 and the wafer identification area 320. If the height of the center point H2 of the upper edge of the actuator identification area 310 is equal to the height of the center point H1 of the lower edge of the wafer identification area 320, that is, the center point H2 of the upper edge of the actuator identification area 310 coincides with the center point H1 of the lower edge of the wafer identification area 320, there is no relative position deviation between the wafer 200 and the end effector 121; if the height of the center point H2 of the upper edge of the actuator identification area 310 is not equal to the height of the center point H1 of the lower edge of the wafer identification area 320, that is, the center point H2 of the upper edge of the actuator identification area 310 is not coincident with the center point H1 of the lower edge of the wafer identification area 320, a relative position deviation occurs between the wafer 200 and the end effector 121, which is also a situation that is easy to occur when the end effector 121 is tilted, and thus it is judged that the position of the end effector 121 is abnormal.

[0056] In one embodiment, see Figure 3 The detection device further includes an alarm 140 connected to the detection component 130. Specifically, the alarm 140 is connected to the position determination module 132. When the detection component 130 determines that the position of the end effector 121 is abnormal, the alarm 140 is triggered to sound an alarm.

[0057] In this embodiment, the alarm 140 works in conjunction with the feedback adjustment module 133. When the feedback adjustment module 133 automatically adjusts the position of the end effector 121, the alarm 140 can be used as an auxiliary means. Once a position abnormality is detected, the alarm 140 will immediately sound an alarm to remind the operator and ensure effective monitoring of abnormal conditions of the robotic arm 120.

[0058] In one embodiment, the detection component 130 is a pre-trained neural network model. The steps of training the neural network model include: collecting image data, including images of the end effector 121 without the wafer 200, images of the end effector 121 and the wafer 200, and images of the end effector 121 in various postures (for example, different tilt angles and horizontal states), etc.; preprocessing the image data, dividing the image data into a training set, a validation set, and a test set; selecting a suitable neural network model, for example, a multi-layer perceptron, a convolutional neural network, or a recurrent neural network, etc. Initialize the weights and biases in the neural network model, select a suitable loss function and optimization algorithm, input the training data into the neural network model to train the neural network model, verify the performance of the neural network model through the validation set, and evaluate the performance of the neural network model on the test set, and finally obtain a trained neural network model.

[0059] The present invention also provides a method for detecting the position of a robot arm, comprising: Using the robot arm position detection device as in any of the above embodiments, the imaging element 110 is disposed on the end effector 121 of the robot arm 120; During the detection process, the camera element 110 collects the detection image 300 of the end effector 121 in real time and sends it to the detection component 130. The detection component 130 processes each detection image 300, identifies the position of the end effector 121 in the detection image 300, and determines whether the position of the end effector 121 is abnormal. If the position of the end effector 121 is determined to be abnormal, the position of the end effector 121 is adjusted until the detection component 130 determines that the position of the end effector 121 is normal.

[0060] In the description of the present invention, it should be understood that the terms "including" and "having" and any variations thereof used herein are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0061] It should be understood that the orientation or position relationship indicated by terms such as "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0062] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0063] Although the embodiments of the present invention are described in detail above, it is obvious to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention in the claims. Moreover, the present invention described herein may have other embodiments and may be implemented or realized in a variety of ways. Unless otherwise defined, the technical terms or scientific terms used herein should be understood by people with ordinary skills in the field to which the present invention belongs.

Claims

1. A robot arm position detection device, characterized in that: include: An imaging element is provided on the part of the robot arm close to the end effector and is used to collect detection images in real time during the process of the end effector taking and placing the wafer; The detection component is used to identify the position of the end effector in the detection image to determine whether the position of the end effector is abnormal.

2. The robot arm position detection device according to claim 1, characterized in that: The detection component comprises: An image recognition module, connected to the imaging element and used to identify an actuator recognition area from the detection image, wherein the actuator recognition area is a minimum rectangular area in the detection image that includes the end effector; The position determination module is connected to the image recognition module and is used to determine whether the position of the end effector is abnormal according to the width and length of the identification area of ​​the effector.

3. The robot arm position detection device according to claim 2, characterized in that: The position determination module is further used to determine whether the actuator identification area satisfies all of the following determination conditions: Determining whether the width of the actuator recognition area is within a set width range; Determining whether the length of the actuator recognition area is within a set length range; If so, it is determined that the position of the end effector is normal; if not, it is determined that the position of the end effector is abnormal.

4. The robot arm position detection device according to claim 3, characterized in that: Also includes: A feedback adjustment module is connected to the position judgment module and the robotic arm and is used to control the robotic arm to adjust the position of the end effector until the width of the actuator recognition area is within the set width range and the length of the actuator recognition area is within the set length range when the position judgment module determines that the position of the end effector is abnormal.

5. The robot arm position detection device according to claim 3, characterized in that: The image recognition module is further used to identify a wafer recognition area from the detection image, where the wafer recognition area is the smallest rectangular area in the detection image that contains the wafer; When taking the wafer, the judgment condition of the actuator identification area also includes: Determine whether a difference between a height distance between a left edge point of an upper edge of the actuator identification region and a left edge point of a lower edge of the wafer identification region and a height distance between a center point of an upper edge of the actuator identification region and a center point of a lower edge of the wafer identification region is within a first error range; Determine whether the difference between the height distance between the right edge point of the upper edge of the actuator identification area and the right edge point of the lower edge of the wafer identification area and the height distance between the center point of the upper edge of the actuator identification area and the center point of the lower edge of the wafer identification area is within the first error range.

6. The robot arm position detection device according to claim 3, characterized in that: The image recognition module is further used to identify a wafer recognition area from the detection image, where the wafer recognition area is the smallest rectangular area in the detection image that contains the wafer; When taking the wafer, the judgment condition of the actuator identification area also includes: Determine whether the difference between the sum of the height of the left edge point of the upper edge of the actuator identification area and the left edge point of the lower edge of the wafer identification area and the sum of the height of the center point of the upper edge of the actuator identification area and the center point of the lower edge of the wafer identification area is within a second error range; Determine whether the difference between the sum of the height of the right edge point of the upper edge of the actuator identification area and the right edge point of the lower edge of the wafer identification area and the sum of the height of the center point of the upper edge of the actuator identification area and the center point of the lower edge of the wafer identification area is within the second error range.

7. The robot arm position detection device according to claim 2, characterized in that: The image recognition module is further used to identify a wafer recognition area from the detection image, where the wafer recognition area is the smallest rectangular area in the detection image that contains the wafer; When placing the wafer, the judgment condition of the actuator identification area also includes: It is determined whether a difference between a height of a center point of an upper edge of the actuator recognition area and a height of a center point of a lower edge of the wafer recognition area is within a third error range.

8. The robot arm position detection device according to any one of claims 1 to 7, characterized in that: It also includes an alarm connected to the detection component, and when the detection component determines that the position of the end effector is abnormal, the alarm is triggered to sound an alarm.

9. The robot arm position detection device according to any one of claims 1 to 7, characterized in that: The detection component is a pre-trained neural network model.

10. A method for detecting the position of a robot arm, characterized in that: include: Using the robot arm position detection device as described in any one of claims 1 to 9, the camera element is arranged on the end effector of the robot arm; During the detection process, the camera element collects the detection image of the end effector in real time and sends it to the detection component. The detection component processes each of the detection images, identifies the position of the end effector in the detection image, so as to determine whether the position of the end effector is abnormal. If the position of the end effector is determined to be abnormal, the position of the end effector is adjusted until the detection component determines that the position of the end effector is normal.

Images