Vehicle charging port positioning method based on visual identification

Abstract

The invention discloses a vehicle charging port positioning method based on visual identification, and relates to the technical field of machine vision. The method comprises the steps that a visual system comprises a laser emitter and a camera, the laser emitter emits laser to be aligned with a charging port, the visual system shoots the charging port, a point cloud picture is generated through intelligent image analysis, and through a series of sitting conversion, pixel coordinates, image coordinates, camera coordinates and mechanical arm coordinates are sequentially subjected to coordinate matching; and calculating to obtain accurate position coordinate information of the charging port. Accurate positioning can be achieved, the unmanned charging technology based on visual identification can monitor and position the charging device and the charging pile in real time through the camera, and it is ensured that the charging gun is accurately aligned with the charging port and inserted into the charging port for charging.

Description

Technical Field

[0001] The present invention relates to the field of machine vision technology, and in particular to a vehicle charging port positioning method based on visual recognition. Background Art

[0002] With the popularization of electric vehicle applications and the advancement of science and technology, automatic charging technology has emerged. The management of electric transport vehicles in today's smart factories and petrochemical enterprises is becoming more and more standardized, and the charging application scenarios of electric vehicles have achieved charging freedom in many fields. The current positioning of vehicle charging ports is usually achieved by acquiring images through image acquisition equipment, and determining the location of the charging port based on the matching of the image with the important features of the target charging port. However, the above method has low positioning coordinate accuracy for visual positioning and image processing, and is easily disturbed by external light, resulting in deviations in recognition, which ultimately leads to large errors in the identified charging port position, and sometimes it cannot be recognized. Summary of the invention

[0003] Therefore, in view of the above technical problems, the present invention needs to provide a vehicle charging port positioning method based on visual recognition. By designing a new algorithm, the coordinate position of the charging port can be accurately calculated, which is convenient for the charging gun to be accurately inserted for charging, with high degree of automation and strong adaptability.

[0004] The technical solution of the present invention is as follows:

[0005] A vehicle charging port positioning method based on visual recognition, characterized in that the method comprises:

[0006] The visual system includes a laser transmitter and a camera. The laser transmitter emits laser light to aim at the vehicle charging port. The visual system takes a picture of the charging port and generates a point cloud map through intelligent image analysis to calculate the accurate position coordinate information of the vehicle charging port:

[0007] S1: Scan and collect to generate point cloud images;

[0008] S2: Image preprocessing to generate model: Using the acquired point cloud data, rasterizing after noise point processing to generate a digital surface model, rasterizing by constructing an irregular triangulated network model to obtain digital elevation model data, or converting it into contour data;

[0009] S3: Coordinate conversion: Perform coordinate conversion in sequence: pixel coordinates - image coordinates - camera coordinates - robotic arm coordinates to determine the location of the charging port.

[0010] The vehicle charging port positioning method based on visual recognition is characterized in that the visual system includes a laser emitter and a camera, the laser emitter and the camera are on the same horizontal line called a baseline, the distance between them is s, the focal length of the camera is f, and the angle between the line connecting the laser emitter and the charging port and the baseline is β.

[0011] The vehicle charging port positioning method based on visual recognition is characterized in that the scanning acquisition generates a point cloud map:

[0012] Assume that the charging port is illuminated by a laser and the position reflected back to the camera imaging plane is point p. The triangle formed by the laser emitter, camera and charging port is similar to the triangle formed by the camera, imaging point p and auxiliary point n. Auxiliary point n is the intersection of the parallel lines connecting the laser emitter and the charging port on the imaging plane. The distance between p and auxiliary point n is pn=x, q is the vertical distance between the charging port and the line connecting the laser emitter and the camera, and d is the distance between the charging port and the laser emitter. From similar triangles, we can get: f / x=q / s, q=fs / x; X is calculated in two parts: X=x1+x2=f / tanβ+pixelSize×position, where pixelSize is the pixel unit size, and position is the position of the imaged pixel coordinate relative to the imaging center. Finally, the distance d can be obtained:

[0013] d=q / sinβ, the point cloud data contains (X, Y, Z) coordinate information, and all the coordinate information is combined into a data structure to obtain a 3D point cloud map of the charging port.

[0014] The vehicle charging port positioning method based on visual recognition is characterized in that the image preprocessing generation model: using the acquired point cloud data, rasterizing it after noise point processing, to generate a digital surface model; at the same time, through automatic preprocessing of the point cloud data, and then using manual editing to perform ground filtering, non-ground points are filtered out, and the remaining ground points are rasterized by constructing an irregular triangulated network model to obtain digital elevation model data, or converted into contour data.

[0015] The vehicle charging port positioning method based on visual recognition is characterized in that the coordinate transformation is:

[0016] The coordinate transformation steps are as follows:

[0017] Pixel coordinate system—image coordinate system—camera coordinate system—robotic arm coordinate system

[0018] The following four coordinate systems are involved:

[0019] uv: pixel coordinate system, unit pixelo-xy: image coordinate system, based on the real plane space of the photosensitive components in the camera, the origin is the midpoint of the imaging plane, unit mm

[0020] Oc-XcYcZc: camera coordinate system, with the optical center as the origin, unit: mm

[0021] Or-XrYrZr: robot coordinate system, describing the position of the robot arm, unit: mmP1: a point on the camera coordinate system, indicating the position of the charging port in the three-dimensional space of the robot arm

[0022] p: Point p is a point in the image coordinate system, its coordinates in the image coordinate system are (x, y) and its coordinates in the pixel coordinate system are (u, v).

[0023] The vehicle charging port positioning method based on visual recognition is characterized in that the pixel coordinate system is converted to the image coordinate system: O-uv is the pixel coordinate system, O-xy is the image coordinate system, to complete the conversion from pixel dimension to mm dimension, it is necessary to know the pixel size dx and dy in the xy direction, the unit is mm, which represents the actual size of a pixel point on the sensor. Assuming the image resolution is 1280x960 and the sensor size is 32mmx24mm, the conversion relationship is

[0024] d x =32mm / 1280

[0025] d y =24mm / 960

[0026] x=(u-640)d x =ud x-640 d x =ud x -u0d x

[0027] y=(v-480)dy=vdy-480dy=vdy-v0dy

[0028] Among them, u0 and v0 are the translation components from the pixel coordinate system to the image coordinate system.

[0029] The vehicle charging port positioning method based on visual recognition is characterized in that the image coordinate system is converted into the camera coordinate system: point p is a point on the image coordinate system, P1 is a point on the camera coordinate system, f is the focal length of the camera, and O to Oc is the distance from the sensor to the optical center, which is calculated by similar triangles:

[0030]

[0031] Conversion relationship:

[0032]

[0033]

[0034] Zc represents the distance between the camera and the charging port, which is directly obtained through the 3D camera and is therefore converted into coordinates on a two-dimensional plane for calculation.

[0035] The vehicle charging port positioning method based on visual recognition is characterized in that the camera coordinate system is converted into the robot coordinate system: r -X r Y r Z r is the robot coordinate system, O c -X c Y c Z c is the camera coordinate system. The coordinates of point P1 in the camera coordinate system are: (xc, yc, zc), the angle between it and the x-axis is a, and the distance to the origin is R. Assuming that the robot coordinate system and the camera coordinate system Z-axis coincide, the robot coordinate system rotates around the Z-axis by a certain angle yaw. The coordinates of point P in the image coordinate system (xr, yr, zr) are calculated as follows:

[0036] xr=R*COS(a+yaw)=R*cosa*cos(yaw)-R*sina*sn(yaw)

[0037] yr=R*sin(a+yaw)=R*sina*cos(yaw)+R*cosa*sin(yaw)

[0038] xc=R*cosa

[0039] yc=R*sina

[0040] So we have:

[0041] Xr=xc*cos(yaw)-yc*sin(yaw)

[0042] Yr=Xc*sin(yaw)+yc*cos(yaw)

[0043] Zr=zc

[0044] Convert to matrix form:

[0045]

[0046] in

[0047] The rotation matrix R of the rotation angle a around the Z axis -1 z.

[0048] The vehicle charging port positioning method based on visual recognition is characterized in that it includes a charging robot, which has the following structure: it includes a base, a central control center and a power supply are installed in the base, a visual display screen is installed on the base, the visual display screen is connected to the signal control of the central control center, a six-axis collaborative robotic arm is also installed on the base, the front end of the six-axis collaborative robotic arm is fixedly connected to a mounting seat, a charging gun is fixedly installed on the mounting seat, and a visual system is installed above the charging gun.

[0049] The vehicle charging port positioning method based on visual recognition is characterized in that the central control center receives the accurate position coordinate information of the charging port, and the robotic arm inserts the charging gun according to the received charging port position coordinate information, and starts charging after successfully completing the insertion of the gun.

[0050] The method of the present invention can realize accurate positioning of the charging port, accurate insertion of the charging gun, and unmanned charging, and has the following characteristics:

[0051] 1. Precise positioning: Unmanned charging technology based on visual recognition can monitor and locate charging equipment and charging piles in real time through cameras, and can accurately locate before contacting the charging pile to ensure precise alignment of the charging interface.

[0052] 2. High flexibility: Compared with other charging technologies, unmanned charging technology based on visual recognition is more flexible. It does not require special physical connections or drive devices, and can complete the charging task using only cameras and image processing software. This makes unmanned charging equipment more flexible in adapting to charging ports of different types and shapes.

[0053] 3. High degree of automation: Unmanned charging technology based on visual recognition can realize a fully automated charging process. The charging device can detect and identify the position and posture of the charging port in real time through image processing algorithms, and automatically adjust the position and angle of the charging device to automatically align the charging port and start charging.

[0054] 4. Real-time monitoring and adjustment: Unmanned charging technology based on visual recognition can monitor the charging process in real time and adjust the charging parameters according to the feedback from the charging port, which makes the charging process more reliable and efficient, and can respond to abnormal situations and failures in a timely manner.

[0055] 5. Strong adaptability: Unmanned charging technology based on visual recognition can adapt to charging ports of different types and shapes, such as electric vehicles, robots, etc. By updating the training model and algorithm, the charging equipment can have a certain learning and adaptability, so as to better meet the needs of different charging ports.

[0056] 6. Accurate positioning: Through intelligent positioning technology, the charging port is accurately positioned, which improves the convenience and accuracy of charging. The charging device can automatically dock with the charging port, reducing the tediousness of manual operation and improving charging efficiency.

[0057] In general, the charging port positioning method based on visual recognition of the present invention has the characteristics of precise positioning, high flexibility, high degree of automation, real-time monitoring and adjustment, and strong adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of the structure of the charging device of the present invention.

[0059] Figure 2 This is a diagram of the locations of the laser emitter and camera.

[0060] Figure 3 This is the image of the charging port taken by the camera.

[0061] Figure 4 Generate the charging port image processed by the model for image preprocessing.

[0062] Figure 5 There are four coordinate systems: image, pixel, camera, and robot.

[0063] Figure 6 is the pixel and image coordinate system.

[0064] Figure 7 is the image and camera coordinate system.

[0065] Figure 8 is the robot and camera coordinate system.

[0066] Fig. 9 The working field of view of the camera. DETAILED DESCRIPTION

[0067] Example 1

[0068] See also Figure 1 .

[0069] The automatic charging system consists of the following components:

[0070] 1. Charging equipment: including charging piles, charging robots, etc.

[0071] 2. Battery Management System (BMS): used to manage and monitor batteries in charging equipment. The system can monitor the status of the battery in real time, such as power, charging speed and charging temperature, and perform charging optimization control through intelligent algorithms to ensure the stability and safety of the charging process.

[0072] Communication network: Through the communication network, charging equipment can communicate remotely with the charging management center or other related equipment to obtain charging strategies, send charging reports, and receive control instructions.

[0073] Central Control Center: As the core part of the unmanned charging system, the control system is used to schedule and execute charging tasks. The system consists of hardware devices and software programs. The hardware devices are responsible for performing charging operations, such as connecting to the power supply, controlling current and voltage, etc. The software programs are responsible for coordinating the entire charging process, including path planning, charging strategy decision-making, monitoring and fault handling, etc.

[0074] The structure of the charging robot is as follows: it includes a base 1, in which a central control center and a power supply are installed, a visual display screen 2 is installed on the base 1, and the visual display screen 2 is connected to the signal control of the central control center. A six-axis collaborative robotic arm 3 is also installed on the base 1, and a mounting seat 4 is fixedly connected to the front end of the six-axis collaborative robotic arm 3, and a charging gun 5 is fixedly installed on the mounting seat 4, and a visual system 6 is installed above the charging gun 5.

[0075] Workflow

[0076] 1. The electric vehicle automatically drives to the designated location and opens the protective cover, and informs the robot to start charging through the background intelligent interconnection.

[0077] 2. The visual system captures the charging port and generates a point cloud through intelligent image analysis. It calculates the accurate coordinate information of the charging port location and sends it to the robotic arm.

[0078] 3. The robotic arm moves and the charging gun is accurately inserted into the charging port.

[0079] Example 2

[0080] A vehicle charging port positioning method based on visual recognition, characterized in that the method comprises:

[0081] The visual system includes a laser transmitter and a camera. The laser transmitter emits laser light to aim at the vehicle charging port. The visual system takes a picture of the charging port and generates a point cloud map through intelligent image analysis to calculate the accurate position coordinate information of the vehicle charging port:

[0082] S1: Scan and collect to generate point cloud images;

[0083] S2: Image preprocessing to generate model: Using the acquired point cloud data, rasterizing after noise point processing to generate a digital surface model, rasterizing by constructing an irregular triangulated network model to obtain digital elevation model data, or converting it into contour data;

[0084] S3: Coordinate conversion: Perform coordinate conversion in sequence: pixel coordinates - image coordinates - camera coordinates - robotic arm coordinates to determine the location of the charging port.

[0085] See also Figure 2 The visual system includes a laser emitter and a camera. The laser emitter and the camera are on the same horizontal line called the baseline. The distance between them is s, the focal length of the camera is f, and the angle between the line connecting the laser emitter and the charging port and the baseline is β.

[0086] 1. Scan and collect to generate point cloud map

[0087] like Figure 2 As shown, the scanning acquisition generates a point cloud map:

[0088] Assume that the charging port is illuminated by a laser and the position reflected back to the camera imaging plane is point p. The triangle formed by the laser emitter, camera and charging port is similar to the triangle formed by the camera, imaging point p and auxiliary point n. Auxiliary point n is the intersection of the parallel lines connecting the laser emitter and the charging port on the imaging plane. The distance between p and auxiliary point n is pn=x, q is the vertical distance between the charging port and the line connecting the laser emitter and the camera, and d is the distance between the charging port and the laser emitter. From similar triangles, we can get: f / x=q / s, q=fs / x; X is calculated in two parts: X=x1+x2=f / tanβ+pixelSize×position, where pixelSize is the pixel unit size, and position is the position of the imaged pixel coordinate relative to the imaging center. Finally, the distance d can be obtained:

[0089] d=q / sinβ, the point cloud data contains (X, Y, Z) coordinate information, and all the coordinate information is combined into a data structure to obtain a 3D point cloud map of the charging port.

[0090] Several 3D point clouds such as Figure 3 shown.

[0091] 2. Image preprocessing generation model

[0092] The image preprocessing generation model described above uses the acquired point cloud data to generate a digital surface model by rasterizing it after noise point processing. At the same time, the point cloud data is automatically preprocessed and then ground filtered by manual editing to filter out non-ground points. The remaining ground points are rasterized by constructing an irregular triangulated network model to obtain digital elevation model data or convert them into contour data.

[0093] After processing Figure 4 shown.

[0094] 3. Coordinate transformation

[0095] The coordinate transformation steps are as follows:

[0096] Pixel coordinate system—image coordinate system—camera coordinate system—robotic arm coordinate system

[0097] Figure 5 k involves the following four coordinate systems:

[0098] uv: pixel coordinate system, unit pixelo-xy: image coordinate system, based on the real plane space of the photosensitive components in the camera, the origin is the midpoint of the imaging plane, unit mm

[0099] Oc-XcYcZc: camera coordinate system, with the optical center as the origin, unit: mm

[0100] Or-XrYrZr: robot coordinate system, describing the position of the robot arm, unit: mmP1: a point on the camera coordinate system, indicating the position of the charging port in the three-dimensional space of the robot arm

[0101] p: Point p is a point in the image coordinate system, its coordinates in the image coordinate system are (x, y) and its coordinates in the pixel coordinate system are (u, v).

[0102] 4. Convert pixel coordinate system to image coordinate system

[0103] See also Figure 6 , pixel coordinate system is converted to image coordinate system: O-uv is the pixel coordinate system, O-xy is the image coordinate system. To complete the conversion from pixel dimension to mm dimension, you need to know the pixel size dx and dy in the xy direction, the unit is mm, which represents the actual size of a pixel on the sensor. Assume that the image resolution is 1280x960 and the sensor size is 32mmx24mm. The conversion relationship

[0104] d x =32mm / 1280

[0105] d y =24mm / 960

[0106] x=(u-640)d x =ud x-640 d x =ud x -u0d x

[0107] y=(v-480)dy=vdy-480dy=v dy-v0dy

[0108] Among them, u0 and v0 are the translation components from the pixel coordinate system to the image coordinate system.

[0109] 5. Convert image coordinate system to camera coordinate system

[0110] See also Figure 7, the image coordinate system is converted to the camera coordinate system: point p is a point on the image coordinate system, P1 is a point on the camera coordinate system, f is the focal length of the camera, and O to Oc is the distance from the sensor to the optical center, which is calculated by similar triangles:

[0111]

[0112] Conversion relationship:

[0113]

[0114]

[0115] Zc represents the distance between the camera and the charging port, which is directly obtained through the 3D camera and is therefore converted into coordinates on a two-dimensional plane for calculation.

[0116] 6. Convert the camera coordinate system to the robot coordinate system

[0117] See also Figure 8 ,

[0118] The camera coordinate system is converted to the robot coordinate system: r -X r Y r Z r is the robot coordinate system, O c -X c Y c Z c is the camera coordinate system. The coordinates of point P1 in the camera coordinate system are: (xc, yc, zc), the angle with the x-axis is a, and the distance to the origin is R. Assuming that the Z-axis of the robot coordinate system coincides with the camera coordinate system, the robot coordinate system rotates around the Z-axis by a certain angle yaw, and the coordinate information of point P in the image coordinate system (xr, yr, zr) is calculated.

[0119] xr=R*COS(a+yaw)=R*cosa*cos(yaw)-R*sina*sn(yaw)

[0120] yr=R*sin(a+yaw)=R*sina*cos(yaw)+R*cosa*sin(yaw)

[0121] xc=R*cosa

[0122] yc=R*sina

[0123] So we have:

[0124] Xr=xc*cos(yaw)-yc*sin(yaw)

[0125] Yr=xc*sin(yaw)+yc*cos(yaw)

[0126] Zr=zc

[0127] Convert to matrix form:

[0128]

[0129] in

[0130] The rotation matrix R of the rotation angle a around the Z axis -1 Z .

[0131] 7. The robot arm inserts the charging gun (accuracy ±0.05mm) according to the received P1 point coordinate information, that is, the charging port position coordinate information, and starts charging after successful completion of the insertion.

[0132] 8. Electric vehicle parking accuracy:

[0133] The vehicle uses RTK positioning and automatic driving technology. After the vehicle stops, the error range can reach ±30cm; within the working range of vision and robotic arms;

[0134] See also Fig. 9 The camera’s working field of view is 995mm*1270mm, which can cover the parking error of the vehicle.

[0135] The working radius of the robotic arm is 1200mm, which can also cover the parking range of the vehicle.

[0136] The movement process of the robotic arm is from point A to point B. Point A is a fixed position, and point B is the charging port position sent by vision (the charging position is different each time).

[0137] The electric vehicle sends a signal that charging is complete, and the robot automatically removes the charging gun.

[0138] The electric vehicle automatically leaves and the entire charging action is completed smoothly.

[0139] The implementation effect of the present invention:

[0140] 1. Intelligence: Through the intelligent interconnection of vehicles, automatic charging robots and charging piles, the entire charging vehicle operation process is unmanned, automated and intelligent; the monitoring and operation and maintenance platform can monitor the status of all equipment in real time.

[0141] 2. Improve charging efficiency: Unmanned charging technology can automatically identify and connect to the charging port without manual operation, greatly saving the user's charging time. This can improve charging efficiency, reduce the time users wait for charging, and make better use of resources when charging facilities are limited.

[0142] 3. Improve charging convenience: Unmanned charging technology makes the charging process more convenient. Users only need to park the vehicle at a designated location, and the system will automatically identify and charge it. This eliminates the tedious steps of finding charging piles and plugging and unplugging charging cables in the traditional way, providing users with a more convenient and pleasant charging experience.

[0143] 4. Reduce personnel input: Unmanned charging technology can realize the charging process in an automated and unmanned way. Compared with the traditional method that requires personnel to perform charging operations, monitoring and maintenance, unmanned charging technology can reduce the demand for human resources, reduce labor costs, and also reduce the potential risks brought by manual operations.

[0144] 5. Improve the utilization rate of charging facilities: Due to the automation and intelligent characteristics of unmanned charging technology, charging facilities can be managed and dispatched more effectively, improving the utilization rate of charging facilities. By real-time monitoring of charging demand and equipment status, the use plan of charging facilities can be optimized, the idle time of charging facilities can be reduced, and the efficiency of resource utilization can be improved.

Claims

1. A vehicle charging port positioning method based on visual recognition, characterized in that: The method includes: The visual system includes a laser transmitter and a camera. The laser transmitter emits laser light to aim at the vehicle charging port. The visual system takes a picture of the charging port and generates a point cloud map through intelligent image analysis to calculate the accurate position coordinate information of the vehicle charging port: S1: Scan and collect to generate point cloud images; S2: Image preprocessing to generate model: Using the acquired point cloud data, rasterizing after noise point processing to generate a digital surface model, rasterizing by constructing an irregular triangulated network model to obtain digital elevation model data, or converting it into contour data; S3: Coordinate conversion: Perform coordinate conversion in sequence: pixel coordinates - image coordinates - camera coordinates - robotic arm coordinates to determine the location of the charging port.

2. The vehicle charging port positioning method based on visual recognition according to claim 1 is characterized in that: The visual system refers to a laser emitter and a camera. The laser emitter and the camera are on the same horizontal line called the baseline. The distance between them is s, the focal length of the camera is f, and the angle between the line connecting the laser emitter and the charging port and the baseline is β.

3. The vehicle charging port positioning method based on visual recognition according to claim 1 or 2, characterized in that: The scanning acquisition generates a point cloud map: Assume that the charging port is illuminated by the laser and the position reflected back to the camera imaging plane is point p. The triangle formed by the laser emitter, camera and charging port is similar to the triangle formed by the camera, imaging point p and auxiliary point n. Auxiliary point n is the intersection of the parallel line connecting the laser emitter and the charging port on the imaging plane. The distance between p and auxiliary point n is pn=x, q is the vertical distance between the charging port and the line connecting the laser emitter and the camera, and d is the distance between the charging port and the laser emitter. From similar triangles, we can get: f / x=q / s, q=fs / x; X is calculated in two parts: X=x1+x2=f / tanβ+pixelSize×position, where pixelSize is the pixel unit size, and position is the position of the imaged pixel coordinate relative to the imaging center. Finally, the distance d can be obtained: d=q / sinβ, the point cloud data contains (X, Y, Z) coordinate information, and all the coordinate information is combined into a data structure to obtain a 3D point cloud map of the charging port.

4. The vehicle charging port positioning method based on visual recognition according to claim 1 or 2, characterized in that: The image preprocessing generation model described above uses the acquired point cloud data to generate a digital surface model by rasterizing it after noise point processing. At the same time, the point cloud data is automatically preprocessed and then ground filtered by manual editing to filter out non-ground points. The remaining ground points are rasterized by constructing an irregular triangulated network model to obtain digital elevation model data or convert them into contour data.

5. The vehicle charging port positioning method based on visual recognition according to claim 1 or 2, characterized in that: Coordinate transformation: The coordinate transformation steps are as follows: Pixel coordinate system—image coordinate system—camera coordinate system—robotic arm coordinate system The following four coordinate systems are involved: uv: pixel coordinate system, unit pixel o-xy: image coordinate system, based on the real plane space of the photosensitive components in the camera, with the origin at the midpoint of the imaging plane, in mm Oc-XcYcZc: camera coordinate system, with the optical center as the origin, unit: mm Or-XrYrZr: robot coordinate system, describing the position of the robot arm, unit: mm P1: A point on the camera coordinate system, which indicates the position of the charging port in the three-dimensional space of the robot arm. P: Point p is a point on the image coordinate system, with coordinates (x, y) in the image coordinate system and (u, v) in the pixel coordinate system.

6. The method for positioning a vehicle charging port based on visual recognition according to claim 5, characterized in that: Pixel coordinate system converted to image coordinate system: O-uv is the pixel coordinate system, O-xy is the image coordinate system. To complete the conversion from pixel dimension to mm dimension, you need to know the pixel size dx and dy in the xy direction, in mm, which represents the actual size of a pixel on the sensor. Assume that the image resolution is 1280x960 and the sensor size is 32mmx24mm. The conversion relationship is d x =32mm / 1280 d y =24mm / 960 x=(u-640)d x =ud x-640 d x =ud x -u0d x y=(v-480)dy=vdy-480dy=v dy-v0dy Among them, u0 and v0 are the translation components from the pixel coordinate system to the image coordinate system.

7. The method for positioning a vehicle charging port based on visual recognition according to claim 5, characterized in that: The image coordinate system is converted to the camera coordinate system: point p is a point on the image coordinate system, P1 is a point on the camera coordinate system, f is the focal length of the camera, and O to Oc is the distance from the sensor to the optical center, which is calculated by similar triangles: Conversion relationship: Zc represents the distance between the camera and the charging port, which is directly obtained through the 3D camera and is therefore converted into coordinates on a two-dimensional plane for calculation.

8. The method for positioning a vehicle charging port based on visual recognition according to claim 5, characterized in that: The camera coordinate system is converted to the robot coordinate system: r -X r Y r Z r is the robot coordinate system, O c -X c Y c Z c is the camera coordinate system. The coordinates of point P1 in the camera coordinate system are: (xc, yc, zc), the angle with the x-axis is a, and the distance to the origin is R. Assuming that the Z-axis of the robot coordinate system coincides with the camera coordinate system, the robot coordinate system rotates around the z-axis by a certain angle yaw, and the coordinate information of point P in the image coordinate system (xr, yr, zr) is calculated.

9. The method for positioning a vehicle charging port based on visual recognition according to claim 1, characterized in that: It includes a charging robot, which has the following structure: a base, a central control center and a power supply are installed in the base, a visual display screen is installed on the base, the visual display screen is connected to the signal control of the central control center, a six-axis collaborative robotic arm is also installed on the base, the front end of the six-axis collaborative robotic arm is fixedly connected to a mounting seat, a charging gun is fixedly installed on the mounting seat, and a visual system is installed above the charging gun.

10. The method for positioning a vehicle charging port based on visual recognition according to claim 1, characterized in that: The central control center receives the accurate position coordinate information of the charging port, and the robotic arm inserts the charging gun according to the received position coordinate information of the charging port, and starts charging after successfully completing the insertion of the gun.

Images