Construction waste assembly line sorting device based on robot vision

Abstract

The invention discloses a construction waste assembly line sorting device based on robot vision, and relates to the technical field of assembly lines. Comprising a positioning table, a plurality of positioning plates are fixed to the upper end of the positioning table, meanwhile, a conveying assembly is arranged at the upper ends of the positioning plates, a mechanical arm body and an image collector are arranged on the two sides of the conveying assembly respectively, a tail end clamping jaw is arranged at the end of the mechanical arm body, and the tail end clamping jaw is arranged at the upper end of the conveying assembly. Meanwhile, a photoelectric sensor is arranged on the side edge of the conveying assembly, and detected construction waste and to-be-detected construction waste are arranged at the upper end of the conveying assembly and are arranged on the two sides of the photoelectric sensor respectively. According to the invention, the full-automatic process from image acquisition, classification and identification to grabbing and sorting of the mechanical arm is realized, and the dependence on manual operation is reduced, so that the working efficiency is improved, and the labor cost is also reduced.

Description

Technical Field

[0001] The invention relates to the technical field of assembly lines, in particular to a construction waste assembly line sorting device based on robot vision. Background Art

[0002] With the acceleration of urbanization, the annual output of construction waste continues to rise. The traditional manual sorting mode has prominent problems such as high labor intensity, low sorting efficiency, and high misclassification rate. In order to improve the recycling rate of resources and reduce environmental load, the use of automated sorting equipment to replace manual labor has become an inevitable trend in the industry. The current international mainstream technology mostly uses laser scanning combined with robotic arms. For example, the garbage sorting robot developed by Finland realizes target recognition through laser 3D modeling and uses Cartesian coordinate robotic arms to perform sorting operations.

[0003] The Chinese utility model patent with announcement number CN212284935U discloses a metal sheet sorting device based on 3D vision, including a sorting robot, the sorting robot includes a sorting free arm, the end of the sorting free arm is provided with an end picker for sorting parts, and also includes a bracket, the bracket includes a free end located above the sorting free arm, the free end is provided with a camera module for image recognition and connected to the sorting robot for communication, and the field of view of the camera module covers the operating track of the end picker. During the sorting operation, the system identifies the area to be sorted, thereby alarming and avoiding abnormal conditions such as missing parts and parts stacking in the current area, ensuring that the sorting operation is carried out in an orderly manner. The cut steel plate parts are transported to the predetermined position without relying on the cutting logistics system; by obtaining the position and posture information of the parts in real time, it can effectively deal with parts with a certain displacement or tilt and warping, making the grasping more accurate and improving the operation efficiency and safety.

[0004] In the actual sorting process, the above-mentioned and similar sorting devices use a robotic arm with a visual system to replace manual sorting of garbage. Although this can improve sorting efficiency and accuracy to a certain extent, the existing technology still has significant technical bottlenecks and challenges, specifically: 1. Existing laser scanning recognition technology is expensive and requires powerful data processing capabilities and computing resources to support its operation. This high cost not only limits the large-scale application of the technology, but also increases the cost of maintenance and operation. In addition, since laser scanning is sensitive to external environmental conditions (such as light, humidity, etc.), it is easy to produce recognition errors, which will affect the accuracy and efficiency of sorting; 2. Currently, most garbage sorting robotic arms on the market adopt a Cartesian structure design. This type of robotic arm has only three degrees of freedom, and all joints are mobile joints, which makes it difficult to adapt to the complex and changing shapes and sizes of construction waste. Especially when dealing with objects with a large long diameter, the end gripper can only take a single posture to grasp, which greatly limits its operating range and flexibility, thereby affecting the sorting efficiency; 3. Usually rely on a single perception mode (such as vision or touch) and lack effective multi-modal fusion mechanism. For example, in an environment with insufficient light or severe occlusion, it is impossible to effectively identify and locate the target by relying solely on visual sensors. Summary of the invention

[0005] The purpose of the present invention is to provide a construction waste assembly line sorting device based on robot vision to solve the problems raised in the above background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a construction waste assembly line sorting device based on robot vision, comprising a positioning table 1, a plurality of positioning plates 10 are fixed to the upper end of the positioning table 1, and two adjacent positioning plates 10 are connected by hinges, and a conveying component 7 is arranged at the upper end of the positioning plate 10, and a mechanical arm body 2 and an image collector 4 are arranged on both sides of the conveying component 7, respectively, and an end clamp 3 is arranged at the end of the mechanical arm body 2, and the end clamp 3 is arranged at the upper end of the conveying component 7, and a photoelectric sensor 20 is arranged on the side of the conveying component 7, and a construction waste 8 to be measured and a construction waste 9 to be measured are arranged at the upper end of the conveying component 7, and the construction waste 8 to be measured and the construction waste 9 to be measured are arranged on both sides of the photoelectric sensor 20 respectively; The robot body 2, the end gripper 3, the image collector 4, the transmission component 7 and the photoelectric sensor 20 are connected through a main controller. The image collector 4 is used to obtain the image and position information of the measured construction waste 8, and the main controller electrically connects the image collector 4 with the robot body 2 and the end gripper 3 through the image and position information. The photoelectric sensor 20 is used to obtain the regional position information of the measured construction waste 8, and the main controller electrically connects the photoelectric sensor 20 and the transmission component 7 through the regional position information.

[0007] Furthermore, the positioning table 1 includes a positioning table body 12, and the plurality of positioning plates 10 are fixed at the upper end of the positioning table body 12. A positioning table drawer 11 is provided at the lower end of the positioning table body 12. A trash can 13 is provided on the upper side of the positioning plate 10, and the trash can 13 is provided on the side of the conveying component 7.

[0008] Furthermore, the image collector 4 includes a depth camera 5 and a camera support frame 6 , the depth camera 5 is arranged on the side of the transmission component 7 through the camera support frame 6 , and the end of the camera support frame 6 is fixed to the upper end of the positioning plate 10 .

[0009] Furthermore, the conveying assembly 7 includes a conveying main frame 17, and a conveying device support frame 21 is fixed on both sides of the lower end of the conveying main frame 17. A driving roller shaft 16 and a driven roller shaft 19 are respectively provided at both ends of the conveying main frame 17. The driving roller shaft 16 and the driven roller shaft 19 are connected by a synchronous belt 18. At the same time, a motor 14 is provided on one side of the lower end of the driving roller shaft 16. The motor 14 is fixed at the junction of the conveying main frame 17 and the conveying device support frame 21, and the motor shaft of the motor 14 is connected to the rotating shaft of the driving roller shaft 16 by a conveyor belt 15.

[0010] Furthermore, the photoelectric sensor 20 is arranged inside the upper end hole of the conveying device support frame 21 on the side away from the active roller shaft 16.

[0011] Furthermore, the image collector 4 is used to obtain the image and position information of the measured construction waste 8, and the main controller electrically connects the image collector 4 with the mechanical arm body 2 and the end clamp 3 through the image and position information, including: S1: Initial image calculation: According to the photographed image of the construction waste obtained by the image collector 4, the waste category corresponding to the construction waste is determined by the YOLO algorithm; S2: coordinate conversion: according to the camera intrinsic parameter matrix and extrinsic parameter matrix corresponding to the construction waste obtained by the image collector 4, the three-dimensional coordinates corresponding to the construction waste are determined, specifically:

[0012] in: is the camera intrinsic parameter matrix, is the rotation matrix, is the translation vector, is the external parameter matrix, is the X-axis coordinate of the point in the two-dimensional image coordinate system obtained after projection transformation, is the Y-axis coordinate of the point in the two-dimensional image coordinate system obtained after projection transformation, is the X-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the Y-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the Z-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation; S3: Obtaining the running trajectory: performing path planning according to the three-dimensional coordinates corresponding to the construction waste, the three-dimensional coordinates of the robot body 2 , and the three-dimensional coordinates of the end gripper 3 .

[0013] Furthermore, path planning is performed, including: S3.1: Determine the rotation angle: According to the three-dimensional coordinates corresponding to the construction waste, the three-dimensional coordinates of the robot body 2, and the three-dimensional coordinates of the end gripper 3, determine the angle change of each joint in the robot arm, specifically:

[0014] in: is the angle of the i-th joint of the robot, is the forward kinematics function, is the X-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the Y-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the Z-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the joint index number of the robot arm; S3.2: Determine the duty cycle of the PWM signal: by changing the angle of each joint in the robot arm, convert the angle of each joint into the duty cycle of the PWM signal, specifically:

[0015] in: is the duty cycle of the PWM signal corresponding to the i-th joint of the robot arm, is the angle of the i-th joint of the robot, is the maximum rotation angle of the driving motor, The joint index number of the robot arm.

[0016] Compared with the prior art, the present invention has the following beneficial effects: First, the present invention obtains the color and depth information of construction waste through a depth camera, and accurately classifies it through the YOLO algorithm. At the same time, it uses image processing technology and inverse kinematics calculation methods to obtain the operating angle of the robotic arm, realizing a fully automated process from image acquisition, classification and recognition to robotic arm grasping and sorting, reducing dependence on manual operation, thereby not only improving work efficiency but also reducing labor costs; Second: The present invention can complete the rapid sorting of a large amount of construction waste in a short time through a six-degree-of-freedom serial robot arm and a path planning algorithm, which significantly improves the work efficiency compared with the traditional manual sorting method; Thirdly, the present invention ensures that any construction waste that is not processed in time will not be ignored through the linkage mechanism between the photoelectric sensor and the main controller, and can pause the conveyor belt in time to reposition and process the missed items, thereby ensuring the integrity of the sorting process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a structural schematic diagram of the construction waste assembly line sorting device in the present invention; Figure 2 It is a structural schematic diagram of the transmission component in the present invention; Figure 3 A schematic diagram of the process of path planning in the present invention; The corresponding parts names of the numbers in the figure are: Among them: 1. Positioning table; 2. Robot arm body; 3. End gripper; 4. Image collector; 5. Depth camera; 6. Camera support frame; 7. Conveying component; 8. Tested construction waste; 9. Construction waste to be tested; 10. Positioning plate; 11. Positioning table drawer; 12. Positioning table body; 13. Trash can; 14. Motor; 15. Conveyor belt; 16. Active roller shaft; 17. Conveying main frame; 18. Synchronous belt; 19. Driven roller shaft; 20. Photoelectric sensor; 21. Conveying device support frame. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0019] refer to Figure 1-Figure 3The present invention provides a construction waste assembly line sorting device based on robot vision, which includes a positioning table 1, a robot body 2, an image collector 4 and a conveying assembly 7. The positioning table 1 includes a positioning table body 12, a positioning table drawer 11 is arranged at the lower end of the positioning table body 12, and a plurality of positioning plates 10 are fixed at the upper end of the positioning table body 12. In this embodiment, the number of positioning plates 10 is set to 11. That is to say, the 11 positioning plates 10 are all fixed at the upper end of the positioning table body 12, and two adjacent positioning plates 10 are connected by hinges. Further, the robot body 2, the image collector 4 and the conveying assembly 7 are all arranged at the upper end of the positioning plate 10. The robot body 2 and the image collector 4 are respectively arranged on both sides of the conveying assembly 7, and the trash can 13 is arranged on the side of the robot body 2. It is worth noting that the robot body 2 is arranged on the side of the image collector 4, that is, the robot body 2 and the image collector 4 are not located on the same horizontal line.

[0020] Furthermore, the end of the robot body 2 is provided with an end clamp 3 for clamping construction waste, and the end clamp 3 is arranged at the upper end of the conveying assembly 7. In this embodiment, the robot body 2 is configured as a six-degree-of-freedom serial robot arm, and the end clamp 3 is an electric servo clamp on the six-degree-of-freedom serial robot arm, which can control the opening and closing of its clamp and the size of the clamping force through a matching driving motor.

[0021] Furthermore, the image collector 4 includes a depth camera 5 and a camera support frame 6. The depth camera 5 is arranged on the side of the transmission component 7 through the camera support frame 6, and the depth camera 5 is hinged to the end of the camera support frame 6, and the end of the camera support frame 6 is fixed to the upper end of the positioning plate 10. At the same time, the depth camera 5 in this embodiment adopts an RGB-D camera, and the height of the camera support frame 6 can be adjusted by a bolt knob, so that images of construction waste taken from multiple angles can be obtained. It is worth noting that the image collector 4 in this embodiment is hand-eye calibrated by a black and white grid calibration plate, so that the depth camera 5 can obtain images of the construction waste 8 to be measured from various angles, thereby ensuring that the construction waste 8 to be measured can be clearly identified in various postures.

[0022] Furthermore, the conveying assembly 7 includes a conveying main frame 17, and a conveying device support frame 21 is fixed on both sides of the lower end of the conveying main frame 17, and a driving roller shaft 16 and a driven roller shaft 19 are respectively arranged at both ends of the conveying main frame 17. The driving roller shaft 16 and the driven roller shaft 19 are connected by a synchronous belt 18, and a motor 14 is arranged on one side of the lower end of the driving roller shaft 16. The motor 14 is fixed at the intersection of the conveying main frame 17 and the conveying device support frame 21, and the motor shaft of the motor 14 is connected to the rotating shaft of the driving roller shaft 16 by a conveyor belt 15. In other words, by starting the motor 14, the driving roller shaft 16 can be driven to rotate, thereby driving the driven roller shaft 19 to rotate, and then the construction waste can be transmitted through the synchronous belt 18.

[0023] Further, a photoelectric sensor 20 is provided on the side of the synchronous belt 18, and the photoelectric sensor 20 is provided at the upper end of the conveying device support frame 21, and the photoelectric sensor 20 is provided inside the upper end hole of the conveying device support frame 21 away from the side of the active roller shaft 16. In the present embodiment, the photoelectric sensor 20 adopts a diffuse reflection photoelectric sensor, which is used to detect the object existing at the target position. Specifically, the synchronous belt 18 in the present embodiment is provided with the construction waste 8 to be tested and the construction waste 9 to be tested, wherein the construction waste 8 to be tested and the construction waste 9 to be tested are respectively provided on both sides of the photoelectric sensor 20, that is, the construction waste is transmitted through the transmission component 7. That is to say, the construction waste 9 to be tested is transmitted from one end of the active roller shaft 16 to the synchronous belt 18, and in the process of the synchronous belt 18 transmitting the construction waste, the image collector 4 will first shoot the construction waste 9 to be tested, so as to obtain the image and position information corresponding to the construction waste 9 to be tested, and convert the construction waste 9 to be tested into the construction waste 8 to be tested. That is to say, the photoelectric sensor 20 is arranged on the side of the robot body 2 , and the image collector 4 and the photoelectric sensor 20 are arranged on both sides of the robot body 2 , respectively.

[0024] It is worth noting that the transmission component 7 and the photoelectric sensor 20 are connected through the main controller. That is to say, once the photoelectric sensor 20 detects the regional position information of the construction waste 8 to be detected, it indicates that the construction waste 8 to be detected has not been grasped and sorted by the robot body 2 and the end gripper 3, that is, there is a phenomenon of missed picking, so the main controller turns off the operation of the motor 14 in the transmission component 7 according to the obtained regional position information, so as to control the operation of the motor 14 and the operation state of the synchronous belt 18.

[0025] Furthermore, the robot body 2, the end gripper 3 and the image collector 4 are connected through the main controller. That is to say, the image collector 4 is used to obtain the image and position information of the construction waste 8 to be tested, and the main controller electrically connects the image collector 4 with the robot body 2 and the end gripper 3 through the image and position information. Specifically, after the construction waste 9 to be tested obtains the corresponding image and position information through the image collector 4, the main controller will control the movement of the robot body 2 and the end gripper 3 according to the obtained image and position information, so as to grasp and sort the construction waste 8 to be tested. The details are as follows: Step S1: initial image calculation. That is, the image (including color and depth) of the construction waste obtained by the image collector 4 is divided into multiple grids by the YOLO algorithm, and the bounding box in each grid and its corresponding category are obtained, so as to determine the waste category to which the construction waste belongs.

[0026] In the specific implementation process, the construction waste 9 to be tested is photographed by an RGB-D camera with a resolution of 640*480 pixels and a frame rate of 30 frames per second, and a photographed image of the construction waste 9 to be tested is obtained. The data volume of each frame of the photographed image is 1920000 bytes. At the same time, the RGB-D camera and the main controller are connected through a USB interface, so that the photographed image can be transmitted in real time.

[0027] Furthermore, in the main controller, the YOLO algorithm is used to perform an initial calculation on the captured image, that is, the captured image is divided into multiple grids, and a bounding box and its corresponding prediction value are set for each grid (in this embodiment, the category of construction waste is set as the prediction value, including but not limited to plastic, wood and bricks). In other words, by obtaining the probability distribution of the prediction value, the prediction value corresponding to the maximum probability distribution is determined, and the category corresponding to the determined prediction value is the category of the construction waste.

[0028] Step S2: coordinate conversion. That is, according to the camera intrinsic parameter matrix and extrinsic parameter matrix corresponding to the construction waste obtained by the image collector 4, the matrix is ​​converted into three-dimensional coordinates through a conversion formula, specifically:

[0029] in: is the camera intrinsic parameter matrix, is the rotation matrix, is the translation vector, is the external parameter matrix, is the X-axis coordinate of the point in the two-dimensional image coordinate system obtained after projection transformation, is the Y-axis coordinate of the point in the two-dimensional image coordinate system obtained after projection transformation, is the X-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the Y-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, It is the Z-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after the projection transformation.

[0030] In the specific implementation process, the camera intrinsic parameter matrix is ​​set to: , the rotation matrix is ​​set to: , the translation vector is set to: . To further explain, the depth corresponding to the two-dimensional point (320,240) is 1m, and its corresponding three-dimensional coordinates are: (0.5m,-0.3m,1m).

[0031] Step S3: Obtain the running trajectory. That is, according to the three-dimensional coordinates obtained in step S2 and the three-dimensional coordinates of the robot body 2 and the three-dimensional coordinates of the end gripper 3, path planning is performed. The details are as follows: Step S3.1: Determine the rotation angle. That is, according to the three-dimensional coordinates obtained in step S2 and the three-dimensional coordinates of the robot body 2 and the three-dimensional coordinates of the end gripper 3, determine the angle change of each joint in the robot, specifically:

[0032] in: is the angle of the i-th joint of the robot, is the forward kinematics function, is the X-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the Y-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, is the Z-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after projection transformation, The joint index number of the robot arm.

[0033] In the process of specific implementation, since a six-degree-of-freedom serial robot arm is selected in this embodiment, according to the three-dimensional coordinates (0.5m, -0.3m, 1m) obtained above, the angles of each joint are: 30°, -45°, 60°, 0°, 90° and -30°.

[0034] Step S3.2: Determine the duty cycle of the PWM signal. That is, convert the angle of each joint obtained in step S3 into the duty cycle of the PWM signal, so that the main controller can send an operation instruction to the robot body 2 and the end gripper 3 through the duty cycle of the PWM signal, so that the robot body 2 and the end gripper 3 can grasp the construction waste and place it in the corresponding trash can 13.

[0035] In this embodiment, the angle of each joint is converted into the duty cycle of the PWM signal, specifically:

[0036] in: is the duty cycle of the PWM signal corresponding to the i-th joint of the robot arm, is the angle of the i-th joint of the robot, is the maximum rotation angle of the driving motor, The joint index number of the robot arm.

[0037] In the specific implementation process, according to the angles of each joint obtained above, the duty cycles of the corresponding PWM signals are: 16.67%, -25%, 33.33%, 0%, 50% and -16.67%. In other words, according to the duty cycle of the obtained PWM signal, the operation command is sent to the robot body 2 and the end gripper 3.

[0038] Specifically, the operation process of the construction waste assembly line sorting device based on robot vision in this embodiment is as follows: Step 1: The construction waste 9 to be tested is transferred from one end of the driving roller shaft 16 to the synchronous belt 18 , and then transferred to one end of the driven roller shaft 19 through the synchronous belt 18 .

[0039] Step 2: The image of the construction waste 9 to be tested is obtained by the depth camera 5. At this time, the construction waste 9 to be tested is converted from the construction waste 9 to the construction waste 8 to be tested during the transmission process.

[0040] Step three: The main controller determines the corresponding duty cycle of the PWM signal and the type of garbage based on the captured image obtained by the depth camera 5, and sends an operation instruction to the robot body 2 and the end gripper 3 according to the duty cycle of the PWM signal, so that the robot body 2 and the end gripper 3 grasp the measured construction waste 8 and put the measured construction waste 8 into the corresponding garbage can 13.

[0041] Step 4: The photoelectric sensor 20 monitors the objects on the synchronous belt 18. When it detects the presence of construction waste on the synchronous belt 18, it indicates that the construction waste 8 under test has not been grasped and placed by the robot body 2 and the end clamp 3 in time. At this time, the main controller will send a stop command to the motor 14 based on the object position information obtained by the photoelectric sensor 20, that is, stop the transportation of construction waste, until the construction waste 8 under test is placed in the corresponding trash can 13, and then the main controller sends a running command to the motor 14 to continue the transportation of construction waste.

[0042] Step 5: Repeat steps 1 to 5 until all the construction waste 9 to be tested is sorted.

[0043] Although 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 the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is limited by the attached embodiments and their equivalents.

Claims

1. A construction waste assembly line sorting device based on robot vision, comprising a positioning table (1), a plurality of positioning plates (10) are fixed on the upper end of the positioning table (1), and two adjacent positioning plates (10) are connected by hinges, and a transmission component (7) is arranged on the upper end of the positioning plate (10), and a mechanical arm body (2) and an image collector (4) are arranged on both sides of the transmission component (7), characterized in that: The end of the mechanical arm body (2) is provided with an end clamp (3), and the end clamp (3) is arranged at the upper end of the transmission component (7). At the same time, a photoelectric sensor (20) is arranged on the side of the transmission component (7). The upper end of the transmission component (7) is provided with construction waste to be tested (8) and construction waste to be tested (9), and the construction waste to be tested (8) and construction waste to be tested (9) are respectively arranged on both sides of the photoelectric sensor (20); The mechanical arm body (2), the end gripper (3), the image collector (4), the transmission component (7) and the photoelectric sensor (20) are connected via a main controller; the image collector (4) is used to obtain the image and position information of the construction waste (8) to be measured, and the main controller electrically connects the image collector (4) with the mechanical arm body (2) and the end gripper (3) via the image and position information; the photoelectric sensor (20) is used to obtain the regional position information of the construction waste (8) to be measured, and the main controller electrically connects the photoelectric sensor (20) with the transmission component (7) via the regional position information.

2. The construction waste assembly line sorting device based on robot vision according to claim 1 is characterized in that: The positioning table (1) comprises a positioning table body (12), a plurality of positioning plates (10) are fixed on the upper end of the positioning table body (12), a positioning table drawer (11) is provided at the lower end of the positioning table body (12), and a trash can (13) is provided on the upper side of the positioning plate (10), and the trash can (13) is provided on the side of the conveying component (7).

3. The construction waste assembly line sorting device based on robot vision according to claim 1 is characterized in that: The image collector (4) comprises a depth camera (5) and a camera support frame (6); the depth camera (5) is arranged on the side of the transmission component (7) through the camera support frame (6), and the end of the camera support frame (6) is fixed to the upper end of the positioning plate (10).

4. The construction waste assembly line sorting device based on robot vision according to claim 1, 2 or 3, characterized in that: The transmission assembly (7) comprises a transmission main frame (17), and a transmission device support frame (21) is fixed on both sides of the lower end of the transmission main frame (17). A driving roller shaft (16) and a driven roller shaft (19) are respectively arranged at both ends of the transmission main frame (17), and the driving roller shaft (16) and the driven roller shaft (19) are connected by a synchronous belt (18). At the same time, a motor (14) is arranged on one side of the lower end of the driving roller shaft (16), and the motor (14) is fixed at the intersection of the transmission main frame (17) and the transmission device support frame (21), and the motor shaft of the motor (14) is connected to the rotating shaft of the driving roller shaft (16) by a transmission belt (15).

5. The construction waste assembly line sorting device based on robot vision according to claim 1 is characterized in that: The photoelectric sensor (20) is arranged inside the upper end hole of the conveying device support frame (21) on the side away from the active roller shaft (16).

6. The construction waste assembly line sorting device based on robot vision according to claim 1 is characterized in that: The image collector (4) is used to obtain images and position information of the construction waste (8) being measured, and the main controller electrically connects the image collector (4) with the mechanical arm body (2) and the end clamp (3) through the images and position information, and comprises: S1: Initial image calculation: According to the photographed image of the construction waste obtained by the image collector 4, the waste category corresponding to the construction waste is determined by the YOLO algorithm; S2: coordinate conversion: according to the camera intrinsic parameter matrix and extrinsic parameter matrix corresponding to the construction waste obtained by the image collector 4, the three-dimensional coordinates corresponding to the construction waste are determined, specifically: Where: K is the camera intrinsic parameter matrix, R is the rotation matrix, t is the translation vector, [R|t] is the extrinsic parameter matrix, u is the X-axis coordinate of the point in the two-dimensional image coordinate system obtained after the projection transformation in the image coordinate system, V is the Y-axis coordinate of the point in the two-dimensional image coordinate system obtained after the projection transformation in the image coordinate system, X is the X-axis coordinate of the point in the two-dimensional image coordinate system obtained after the projection transformation in the world coordinate system, Y is the Y-axis coordinate of the point in the two-dimensional image coordinate system obtained after the projection transformation in the world coordinate system, and Z is the Z-axis coordinate of the point in the two-dimensional image coordinate system obtained after the projection transformation in the world coordinate system; S3: Obtaining the running trajectory: performing path planning according to the three-dimensional coordinates corresponding to the construction waste, the three-dimensional coordinates of the robot body 2 , and the three-dimensional coordinates of the end gripper 3 .

7. The construction waste assembly line sorting device based on robot vision according to claim 6 is characterized in that: Perform path planning, including: S3.1: Determine the rotation angle: According to the three-dimensional coordinates corresponding to the construction waste, the three-dimensional coordinates of the robot body 2, and the three-dimensional coordinates of the end gripper 3, determine the angle change of each joint in the robot arm, specifically: θ i =f(X,Y,Z) Where: θ i is the angle of the i-th joint of the robot arm, f(X, Y, Z) is the forward kinematics function, X is the X-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after the projection transformation, Y is the Y-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after the projection transformation, Z is the Z-axis coordinate of the point in the two-dimensional image coordinate system in the world coordinate system after the projection transformation, and i is the joint index number of the robot arm; S3.2: Determine the duty cycle of the PWM signal: by changing the angle of each joint in the robot arm, convert the angle of each joint into the duty cycle of the PWM signal, specifically: Where: D i is the duty cycle of the PWM signal corresponding to the i-th joint of the robot arm, θ i is the angle of the i-th joint of the robot arm, θ max is the maximum rotation angle of the driving motor, and i is the joint index number of the robot arm.

Images