An ai vision safety system for a robotic laser cutting workstation

Abstract

The utility model relates to the field of automobile parts robot laser cutting technology safety protection, especially a kind of AI vision safety system for robot laser cutting workstation, including the laser cutting head and its laser power supply, workbench and enclosure being set on laser cutting robot, still include AI vision camera, for the image information of real-time acquisition personnel activity and set article in workstation;Image data processing unit is used to receive and process the image information collected by AI vision camera;Central control unit is connected with image data processing unit, for receiving the information of identification image data processing unit and generating control instruction;Safety execution unit is used to receive control instruction and execute safety measure.The utility model provides a kind of AI vision safety system for robot laser cutting workstation, to solve the safety protection of prior art robot laser cutting vision safety system exists visual blind area, the problem of low work efficiency.

Description

Technical Field

[0001] This utility model relates to the field of safety protection for robotic laser cutting technology of automotive parts, and in particular to an AI vision safety system for robotic laser cutting workstations. Background Technology

[0002] In robotic laser cutting workstations, the movement of the robot body and laser cutting head during the cutting process, as well as the laser emitted during cutting, are characterized by high energy and high risk.

[0003] Currently, traditional safety protection methods mainly rely on physical safety barriers, light curtain sensors, and infrared beam gratings. While physical safety barriers can prevent personnel from entering dangerous areas to some extent, they lack flexibility. Frequent opening and closing of the barriers during operations such as loading and unloading workpieces can reduce work efficiency. Light curtain sensors can only detect obstructions within a specific plane, and their detection capabilities for complex movements and irregularly shaped objects are limited, making it difficult to accurately identify dangerous human behavior and potential safety hazards. Infrared beam gratings are easily interfered with by smoke and debris generated during cutting, frequently resulting in false alarms or missed alarms. They struggle to accurately distinguish between personnel and non-dangerous foreign objects, leading to low reliability and failing to meet the high safety and intelligent protection requirements of modern industrial production.

[0004] In summary, the existing technology has the following defects: (1) There are blind spots in safety protection, and it is impossible to comprehensively and in real time monitor various dangerous situations in the laser cutting workstation. It is difficult to effectively detect some non-linear moving objects and objects and personnel entering from non-light curtain detection planes. (2) The level of intelligence is low. It cannot accurately classify and judge the detected objects, and it is easy to make misjudgments and omissions. It is impossible to take corresponding measures according to different dangerous situations. (3) It affects work efficiency: While ensuring safety, the traditional protection method reduces the operational flexibility and work efficiency of the workstation. Utility Model Content

[0005] This invention provides an AI vision safety system for robotic laser cutting workstations to address the problems of blind spots and low work efficiency in existing robotic laser cutting vision safety systems.

[0006] This utility model is achieved through the following technical solution, providing an AI vision safety system for a robotic laser cutting workstation, including a laser cutting head and its laser power supply, a worktable, and a perimeter enclosure mounted on a laser cutting robot. It also includes: AI vision cameras for real-time acquisition of image information of personnel activities and objects placed within the workstation; at least two AI vision cameras are provided, placed diagonally around the perimeter of the workstation and mounted on the perimeter enclosure; an image data processing unit connected to the AI ​​vision cameras for receiving and processing the image information acquired by the AI ​​vision cameras; the image data processing unit, in collaboration with the AI ​​vision cameras, encloses a virtual monitoring area within the workstation, the boundary coordinates of which are pre-stored in the memory of the image data processing unit, and the boundary of the virtual monitoring area is circular. The system comprises: a circular boundary, defined as a circle with the laser cutting robot as its center and a workbench positioned around the perimeter of the workstation as its radius; a central control unit, connected to an image data processing unit, for receiving and recognizing information from the image data processing unit and generating control commands; and a safety execution unit, connected to the relay output interface of the central control unit, for receiving control commands and executing safety measures. The safety execution unit includes an audible and visual alarm device, a status indicator device, a laser power cut-off module, and a laser cutting robot control module. The laser power cut-off module is connected to the main laser power circuit via an electromagnetic relay for cutting off the laser power supply. The laser cutting robot control module controls the movement of the laser cutting robot.

[0007] Specifically, the AI ​​vision camera is set at a height of 2.5m above the ground, with a lens downward angle of 15°-20°.

[0008] Specifically, the AI ​​vision camera collects in real time the location coordinates of personnel, their postures, whether they are wearing safety helmets, the placement of workpieces, and the distribution and movement trajectory of logistics vehicles.

[0009] Specifically, the AI ​​vision camera is equipped with a grayscale stretching preprocessing module.

[0010] Specifically, the audible and visual alarm device is installed on the fence.

[0011] Specifically, the status indicator device is mounted on the workbench.

[0012] Specifically, the status indicator is a three-color LED screen.

[0013] Specifically, the central control unit is equipped with a human-machine interface.

[0014] Compared with the prior art, the present invention has the following beneficial effects:

[0015] (1) By setting up at least two AI vision cameras, the entire workstation area can be cross-covered without monitoring blind spots. The dynamics of people and objects can be captured in real time, overcoming the defects of traditional local monitoring and adapting to the dynamic dangerous area monitoring when switching between multiple workstations.

[0016] (2) There is no need to frequently operate the physical guardrail. Operators can complete pre-installation of workpieces and cleaning of waste materials outside the virtual monitoring area. The operation is only interrupted when a high risk is confirmed. Compared with the traditional protection method, the daily operation efficiency has been greatly improved. It takes into account both safety and operational flexibility and is suitable for multi-variety small-batch production needs. Attached Figure Description

[0017] Figure 1 This is a system diagram of an AI vision safety system for a robotic laser cutting workstation according to this utility model;

[0018] Figure 2 This is a schematic diagram of the main body of an AI vision safety system for a robotic laser cutting workstation, according to this utility model.

[0019] In the attached diagram: 1. First workbench; 2. Left side enclosure; 3. Laser cutting robot; 4. Virtual monitoring area; 5. Second workbench; 6. First AI vision camera; 7. Rear side enclosure; 8. Intruder; 9. Laser cutting head; 10. Third workbench; 11. Second AI vision camera; 12. Right side enclosure; 13. Operating station. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature; in the description of this application, unless otherwise stated, "multiple" means two or more.

[0022] Please see Figure 1 and Figure 2An AI vision safety system for a robotic laser cutting workstation includes a laser cutting head 9 and its laser power supply, a worktable, a barrier, an AI vision camera, an image data processing unit, a central control unit, and a safety execution unit mounted on a laser cutting robot 3.

[0023] AI vision cameras collect real-time image information of personnel activities and object placement within the workstation. At least two AI vision cameras are installed, placed diagonally around the perimeter of the workstation and mounted on the perimeter fencing. An image data processing unit connects to the AI ​​vision cameras via a network switch to receive and process the image information collected by the cameras. It can connect to an external display for human-machine collaborative setup. The image data processing unit, in conjunction with the AI ​​vision cameras, demarcates a virtual monitoring area 4 within the workstation. The boundary coordinates of the virtual monitoring area 4 are pre-stored in the image data processing unit's memory. The boundary of the virtual monitoring area 4 is circular, with the laser cutting robot 3 as the center and the workbench located on the perimeter of the workstation as the radius, ensuring coverage of the entire work area. The workbench is positioned between the fencing and the laser cutting robot 3. A central control unit connects to the image data processing unit to receive and recognize information from the image data processing unit and generate control commands.

[0024] The safety execution unit is connected to the relay output interface of the central control unit to receive control commands and execute safety measures. The safety execution unit includes an audible and visual alarm device, a status indicator device, a laser power cut-off module, and a laser cutting robot 3 control module. The laser power cut-off module is connected to the main circuit of the laser power supply through an electromagnetic relay to cut off the laser power supply. The laser cutting robot 3 control module is used to control the movement of the laser cutting robot 3.

[0025] This system connects its various units through standardized interfaces to form a closed-loop system of "image acquisition - data processing - decision control - safe execution," ensuring timely identification and rapid response to dangerous situations.

[0026] In this embodiment, the actual size of the workstation is a standard 4000mm × 4000mm. It is equipped with three workbenches: the first workbench 1, the second workbench 5, and the third workbench 10. An operating station 13 is provided, which can be operated by one person. The three workbenches work independently according to different product models, but not simultaneously. The workbenches are arranged in an inverted triangle. Three enclosures are set between each pair of workbenches: the left enclosure 2, the right enclosure 12, and the rear enclosure 7. Two AI vision cameras are installed: the first AI vision camera 6 and the second AI vision camera 11, which can cross and cover the entire working area. They are set on the rear enclosure 7 and the right enclosure 12 at the upper left and lower right corners of the workstation, respectively.

[0027] The image data processing unit and central control unit are located in the robot laser cutting control cabinet, which is situated on the periphery of the rear panel. Two AI vision cameras cross-cover the entire workstation area, eliminating blind spots and capturing the dynamics of personnel and objects in real time, overcoming the limitations of traditional localized monitoring.

[0028] Specifically, the AI ​​vision camera is positioned 2.5m above the ground, with a lens tilt angle of 15°-20°, ensuring a cross-coverage rate of ≥20%, fully covering the three workbench operating areas and a surrounding 500mm buffer zone. The AI ​​vision camera is an industrial-grade, highly adaptable camera with high resolution (≥2 megapixels, 1920×1080), a wide horizontal field of view (≥60°), a vertical field of view (≥34°), good low-light performance, ISO sensitivity (≥12800), and wide dynamic range (WDR) (≥120dB). It can clearly image under strong laser cutting light, spot brightness (≥10000cd / m²), shadows, and dusty environments. It has an IP67 protection rating, withstands temperatures from -10℃ to 60℃, and is suitable for complex industrial workshop conditions. It collects image information within the workstation in real time at a frame rate of ≥30fps, and can capture the rapid movement trajectory of intruders, such as a walking speed of 1.5m per second.

[0029] Specifically, the data collected includes: personnel location coordinates, personnel action postures (stepping, bending over, standing still), personnel wearing safety helmets, workpiece placement position and whether it is offset, and the distribution and movement trajectory of logistics carts and tools; the built-in grayscale stretching preprocessing module initially filters dust interference to ensure the validity of image data.

[0030] The image data processing unit is connected to two AI vision cameras via a data cable with a transmission rate of ≥1000Mbps and a length of ≤20m to receive real-time image data. It is connected to the central control unit via a bus interface with a communication latency of ≤50ms and an anti-interference level of ≥15kV ESD to transmit the processed recognition and analysis results. The image data processing unit employs a median filtering algorithm with a 3×3 window size to remove smoke and dust noise. Histogram equalization enhances image contrast, improving target contour clarity by ≥40%, providing high-quality data for subsequent recognition. Based on the YOLOv8-tiny lightweight AI algorithm model, it classifies and identifies objects in the image, accurately distinguishing four types of targets: "personnel (wearing / not wearing a hat), workpieces, logistics carts, and miscellaneous objects," with an accuracy rate ≥99% and a response time ≤50ms. Based on OpenPose algorithm optimization, it analyzes whether personnel actions constitute dangerous behaviors, including: operators entering the laser cutting area to load / unload / place products, non-operators / logistics carts entering the cutting area, personnel staying in dangerous areas for ≥3 seconds, and operating without a safety helmet, with a behavior judgment accuracy rate ≥98%. The virtual monitoring area 4 is determined based on the workstation's reference coordinates.

[0031] The central control unit connects to the image data processing unit via a bus with a transmission rate of ≥100Mbps and a cycle of ≤1ms to receive target recognition and hazardous behavior analysis results; it also connects to the safety execution unit via a relay output interface to send control commands; a reserved interface supports integration with existing external control systems of the workstation. The laser power supply cut-off module uses an HH52P electromagnetic relay to control the main circuit of the laser power supply, with a response time of ≤20ms. After cut-off, the robot laser cutting control cabinet reset button must be pressed manually to restore operation and prevent accidental restart.

[0032] Specifically, the audible and visual alarm devices are installed on the fence, and are set to yellow and red lights depending on the level of risk.

[0033] Specifically, the status indicator is set on the workbench.

[0034] Specifically, the status indicator is a three-color LED screen. The three colors are: solid green for normal operation, flashing yellow for low-risk warning, and flashing red for medium / high-risk alarm, facilitating remote observation of the system status.

[0035] Specifically, the central control unit is equipped with a human-machine interface.

[0036] Work process:

[0037] During the startup phase, after the workstation is powered on, the system automatically performs a self-test process (AI vision camera → image data processing unit → central control unit → safety execution unit). The self-test takes ≤30 seconds. If it is normal, the three-color LED screen will display a solid green light.

[0038] During operation, the AI ​​vision camera continuously captures images at a frame rate of 30fps and transmits them to the image data processing unit via a data cable. After preprocessing the images, the image data processing unit performs target recognition and behavior analysis, sending the results to the central control unit every 100ms. The central control unit makes decisions in real time. If there is no risk, it maintains normal operation. If a risk is detected, it immediately sends a control command. After receiving the command, the safety execution unit executes the corresponding measures (alarm, pause, power cut-off) within 100ms.

[0039] When the workstation is operating normally, personnel must remain outside the virtual monitoring area set up by the image data processing unit and the AI ​​vision camera to ensure safety.

[0040] When the laser cutting robot 3 is working, if an operator, other personnel, or a logistics vehicle enters the virtual monitoring area, if the operator places materials incorrectly, or if the operator enters the virtual monitoring area without wearing a safety helmet to load or unload materials when the robot stops after laser cutting, a tiered response mechanism will be established based on the analysis results received by the central control unit.

[0041] When it is detected that personnel are approaching the virtual monitoring area without intruding, and the personnel are within 30cm of the fence, it is determined to be low risk, and the operation will not be interrupted. The three-color LED screen will light up yellow.

[0042] When the detected inverted limb intrudes within 10cm or the workpiece offset is ≥10cm, it is judged as medium risk, triggering the safety execution unit to stop command, and the three-color LED screen lights up red and flashes;

[0043] When personnel or vehicles are detected to have completely intruded or are not wearing safety helmets, it is determined to be a high-risk situation. The safety execution unit triggers an emergency braking command to cut off the laser power supply and robot control power supply. The three-color LED screen lights up red and flashes to ensure precise and efficient handling of hazards.

[0044] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. This application is not limited to the exact structures described above and illustrated in the accompanying drawings, and it should not be considered that the specific implementation of this application is limited to these descriptions. For those skilled in the art, various changes and modifications made without departing from the concept of this application should be considered to fall within the protection scope of this application.

Claims

1. An AI vision safety system for a robotic laser cutting workstation, comprising a laser cutting head and its laser power supply mounted on a laser cutting robot, a worktable, and a barrier, characterized in that, include: AI vision cameras are used to collect image information of personnel activities and objects set up in the workstation in real time. At least two AI vision cameras are set up and placed diagonally around the perimeter of the workstation. The AI ​​vision cameras are installed on the fence set up around the perimeter of the workstation. An image data processing unit, which is connected to an AI vision camera, is used to receive and process image information acquired by the AI ​​vision camera; The image data processing unit works in conjunction with the AI ​​vision camera to enclose a virtual monitoring area within the workstation. The boundary coordinates of the virtual monitoring area are pre-stored in the memory of the image data processing unit. The boundary of the virtual monitoring area is a circular boundary, which is a circle with the laser cutting robot as the center and the workbench set around the perimeter of the workstation as the radius. The workbench is set between the enclosure and the laser cutting robot. A central control unit, which is connected to an image data processing unit, is used to receive information from the image data processing unit and generate control commands. A safety execution unit is connected to the relay output interface of the central control unit. It is used to receive control commands and execute safety measures. The safety execution unit includes an audible and visual alarm device, a status indicator device, a laser power supply cut-off module, and a laser cutting robot control module. The laser power supply cut-off module is connected to the main circuit of the laser power supply through an electromagnetic relay and is used to cut off the laser power supply. The laser cutting robot control module is used to control the movement of the laser cutting robot.

2. The AI ​​vision safety system for a robotic laser cutting workstation according to claim 1, characterized in that: The AI ​​vision camera is positioned 2.5m above the ground, with a lens tilt angle of 15°-20°.

3. The AI ​​vision safety system for a robotic laser cutting workstation according to claim 1, characterized in that: The AI ​​vision camera is used to collect real-time data on personnel location coordinates, personnel postures, whether personnel are wearing safety helmets, the placement of workpieces, and the distribution and movement trajectory of logistics vehicles.

4. The AI ​​vision safety system for a robotic laser cutting workstation according to claim 1, characterized in that: The AI ​​vision camera is equipped with a grayscale stretching preprocessing module.

5. The AI ​​vision safety system for a robotic laser cutting workstation according to claim 1, characterized in that: The audible and visual alarm device is installed on the fence.

6. The AI ​​vision safety system for a robotic laser cutting workstation according to claim 1, characterized in that: The status indicator is mounted on the workbench.

7. The AI ​​vision safety system for a robotic laser cutting workstation according to claim 1 or 6, characterized in that: The status indicator is a three-color LED screen.

8. The AI ​​vision safety system for a robotic laser cutting workstation according to claim 1, characterized in that: The central control unit is equipped with a human-machine interface.

Images