Vision-based four-wheel positioning real-time attitude detection device, system and method

Abstract

The invention belongs to the technical field of vehicle four-wheel positioning detection, and particularly relates to a vision-based four-wheel positioning real-time attitude detection device, system and method, and the specific method comprises the steps: carrying out the calibration of a coordinate system, defining the coordinate system of a calibration frame on a test platform, enabling each calibration plate on the calibration frame to correspond to one test device, and carrying out the calibration of the calibration frame; calibrating to obtain a conversion matrix; a vehicle is driven to a test platform, then wheels are driven to rotate, synchronous test equipment starts measurement, toe-in direction angles, inner and outer inclination angles and wheel centers of tires are finally calculated through camera image acquisition, calculation and coordinate conversion, and results of the four wheels are synchronously uploaded to a control storage terminal. And controlling the storage terminal to obtain result information at each moment through post-processing calculation, displaying the result information on an interface in real time and storing the result information in a database. According to the invention, the vehicle is not damaged, dynamic real-time measurement can be realized, and the result is more accurate.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle four-wheel alignment detection, and in particular relates to a vision-based four-wheel alignment real-time posture detection device, system and method. Background Art

[0002] Four-wheel alignment is a core step in vehicle delivery, repair, and maintenance. Through precise alignment and adjustment, it can effectively optimize vehicle handling, enhance driving stability, reduce tire wear, and improve fuel efficiency. It is also a crucial measure to ensure safe driving and extend tire life. In the current technological landscape, the accuracy and reliability of four-wheel alignment devices and methods play a key role in optimizing vehicle driving performance.

[0003] Existing wheel alignment technology primarily relies on wheel alignment sensors, which can be categorized into contact and non-contact types. While mature, contact wheel alignment sensors exhibit numerous deficiencies in practical applications. For example, due to their need for direct contact with the tire or rim, their measurement accuracy is low, and the calibration process requires extensive manual intervention, increasing operational difficulty and workload while also posing the potential risk of scratching or damaging the rim. Furthermore, the high complexity of daily maintenance associated with contact sensors makes them difficult to meet the modern automotive industry's demand for efficiency and convenience.

[0004] In contrast, non-contact wheel alignment sensors have gradually become the mainstream in the market, mainly including static aligners and dynamic aligners. However, these two types of sensors still have obvious limitations:

[0005] Static locators are easily affected by external environmental interference during operation, especially the problem of inclination gain, which is difficult to compensate. At the same time, the vehicle body posture has a greater impact on the toe angle, resulting in low measurement accuracy. In addition, its function is single and cannot meet the diverse needs in complex scenarios.

[0006] Although dynamic locators can obtain vehicle motion status in real time, their system complexity is relatively high, with strict requirements on equipment performance and computing power. They are also relatively expensive and difficult to be widely used in the mid- and low-end markets.

[0007] In summary, existing four-wheel alignment technology still leaves much room for improvement in terms of measurement accuracy, real-time performance, stability, and applicability. To address these technical bottlenecks, this paper proposes a vision-based real-time posture detection device, system, and method for four-wheel alignment. This technology innovatively utilizes computer vision and image processing techniques, combined with advanced algorithm design, to significantly improve measurement accuracy and real-time performance while reducing system complexity and maintenance costs, thus providing a novel solution for the four-wheel alignment field. Summary of the Invention

[0008] The purpose of the present invention is to provide a vision-based four-wheel alignment real-time posture detection device, system and method, which can enable four-wheel alignment to achieve dynamic, real-time and high-precision positioning.

[0009] The technical solutions adopted by the present invention are as follows:

[0010] A vision-based four-wheel positioning real-time posture detection device includes a test platform and a control storage terminal. First slide rails are symmetrically provided on both sides of the test platform, a first slider is movably provided on the first slide rail, rollers are symmetrically and movably connected to both ends of the first slider, a vertical rail is connected at the center of the first slider, a second slider is movably provided on the vertical rail, and test equipment is installed at the end of the second slider. Four test devices are electrically connected to the control storage terminal.

[0011] Preferably, a distance sensor is provided at the inner end of the first slide rail and the inner end of the vertical rail, a limiting mechanism is provided inside the first slide rail and on the inner and upper parts of the vertical rail, the testing equipment includes an equipment body, a light source is provided at the center position of the side of the equipment body, two acquisition cameras are symmetrically provided on the side of the equipment body, the acquisition cameras are symmetrically distributed about the light source, and each group of two acquisition cameras are respectively marked as an upper camera and a lower camera.

[0012] A vision-based four-wheel alignment real-time posture detection system, characterized in that: the detection system is used to control a four-wheel alignment real-time posture detection device, the four-wheel alignment real-time posture detection system includes a sensor acquisition unit and a sensor calculation unit; the sensor acquisition unit is electrically connected to the sensor calculation unit;

[0013] The acquisition unit synchronously acquires the wheel area image at the current moment through the four test devices to obtain acquisition data. The sensor calculation unit performs digital processing based on the visual image according to the acquisition data, and finally obtains the tire's toe direction angle, camber angle and wheel center, and uploads the information to the control storage terminal.

[0014] A vision-based four-wheel alignment real-time posture detection method, the four-wheel alignment real-time posture detection method is a method for using a detection system; the four-wheel alignment real-time posture detection method comprises the following steps:

[0015] Step 1: Perform coordinate system calibration and define the coordinate system of the calibration frame on the test platform. Each calibration plate on the calibration frame corresponds to a test device. Perform coordinate system transformation based on the true value and measured value of the cylindrical surface. Finally, calculate the basic transformation matrix T0 from the device coordinate system to the test platform coordinate system.

[0016] Step 2: Adjust the spacing between the two sets of the first sliders based on the wheelbase of the vehicle to be tested so that the wheels of the test vehicle fit on the floating plates of the two sets of rollers. Then adjust the height of the second slider on the vertical rail so that the test equipment is aligned with the center of the wheel. Based on the adjustment information sent by the control terminal, the test equipment calculates T0 and obtains the current transformation matrix T.

[0017] Step 3: The control storage terminal sends instructions to the synchronizer of the test platform and sensor. The roller drives the wheel to rotate. The synchronizer sends instructions to the four test devices, and the test devices begin measuring. After collecting camera images and calculating point cloud data, it is converted according to the matrix T coordinate system, and then the toe angle, camber angle and wheel center of the four wheels are calculated. The results are then uploaded to the control storage terminal synchronously.

[0018] Step 4: Control the storage terminal to perform post-processing operations based on the data obtained by the test equipment, and calculate the front wheel toe, rear wheel camber, kingpin caster, kingpin inclination and other information based on the tire parameters, display them on the interface in real time and save them in the database.

[0019] Preferably, the specific steps of step 4 are as follows:

[0020] Step 401: According to the instructions of the synchronizer controlling the storage terminal, the four test devices synchronously acquire the wheel area image at the current moment. The sensor computing unit performs image preprocessing based on the acquired data, extracts the sub-pixel centerline of the laser line in the image, and calculates the wheel hub edge feature points.

[0021] Step 402: The wheel hub diameter and tire height are then calculated based on the uploaded tire parameters and converted to a pixel coordinate system as conditional parameters. RANSAC ellipse fitting based on edge feature points is used to locate the wheel hub edge in the image. The tire outer edge ellipse is obtained based on the tire height pixel information. The area between the two concentric ellipses is the tire tread area. The centerline outside the tread area is filtered to obtain the tire centerline information in the image.

[0022] Step 403: Based on the centerline pixel coordinates of the upper and lower cameras and the camera's extrinsic parameters, encoding-based feature point matching, filtering, and restoration operations are performed to obtain tire point cloud information. The tire point cloud is then converted to the test platform coordinate system based on the calibration transformation matrix T.

[0023] Step 404: Finally, a three-dimensional elliptical ring model point cloud template is established based on the tire parameters. The tire point cloud is template matched based on a nonlinear optimization method. The sum of the three-dimensional distances between the three-dimensional points of the tire and the template is minimized through the designed objective function. The optimal solution is obtained to finally obtain the tire's toe angle, camber angle, and wheel center.

[0024] Preferably, the three-dimensional elliptical ring model constructed in step 404 is defined as follows: Let the center of a circle on the XOY plane be O r An ellipse with one axis passing through the X axis and an axis length of 2r, O r The distance from the origin O is R. The torus formed by rotating around the Z axis with O as the control point is a three-dimensional elliptical torus. Let its model formula be:

[0025]

[0026] 0≤v≤π (left) or -π≤v≤0 (right), 0≤u≤2π, -5≤k≤5, The objective function is:

[0027]

[0028] Preferably, in step 404, the point set on the tire is {x i 、y i 、z i │i=0,1,…,n}, the wheel center is (C x ,C y ,C z ), the toe angle is α, the inclination angle is γ, and the angle of rotation around the Y axis is θ. The optimization steps are:

[0029] Step 4041: Initialize the parameters to be optimized; the parameters to be optimized are C x ,C y ,C z ,α,γ,R,r,k,θ, firstly perform circle fitting based on the midpoint of each dense point line on the tire point cloud converted in step 403 to obtain the initial center point and angle C x ,C y ,C z ,α,γ, set the initial value of θ to 0, and then calculate the initial value of R,r,k according to the obtained tire size parameters;

[0030] Step 4042: Then according to C x ,C y ,C z ,α,γ calculate the rotation matrix A∈R ∧ (4×4), transform the tire point cloud coordinate system to the model coordinate system;

[0031] Step 4043: Then, the final objective function is obtained according to formula (1) and formula (2), and the optimal solution of the parameters to be optimized is obtained by constructing a nonlinear optimization objective function;

[0032] Step 4044: Finally, C x ,C y ,Cz ,α,γ are uploaded to the control storage terminal.

[0033] Preferably, the control storage terminal synchronously receives the result information of the four test devices, and calculates the body angle at the current moment based on the four wheel centers. At this moment, the toe angle of each wheel is equal to the toe direction angle minus the body angle; the acquisition and calculation frequency of the four test devices is 10Hz, and the wheel speed is 2s / cycle. The control storage terminal receives and calculates 20 consecutive frames of data in the above-mentioned manner and puts them into the cache, calculates the average value of the 20 frames of toe angles and camber angles as the angle posture at the moment, and calculates the vehicle's front wheel toe, rear wheel camber, kingpin caster angle, and kingpin inclination angle information in combination with the tire parameters, and displays the result information on the interface; the control storage terminal calculates each parameter by updating the cache, displays it on the interface in real time, and writes the information into the database in real time for subsequent inspection.

[0034] The technical effects achieved by the present invention are:

[0035] In the present invention, according to the instructions of the synchronizer, four test devices synchronously acquire images of the wheel area at the current moment. The sensor computing unit performs image preprocessing based on the acquired data, extracts the sub-pixel centerline of the laser line in the image, and calculates the wheel hub edge feature points. The wheel hub diameter and tire height are then calculated based on the uploaded tire parameters and converted to a pixel coordinate system as conditional parameters. Combined with RANSAC ellipse fitting based on edge feature points, the image wheel hub edge is located. The tire outer edge ellipse is obtained based on the tire height pixel information. The area between the two concentric ellipses is the tire tread area. The centerline outside the tread area is filtered to obtain the tire centerline information in the image. Based on the centerline pixel coordinates of the upper and lower cameras and the camera's extrinsic parameters, encoding-based feature point matching, filtering, and restoration operations are performed to obtain tire point cloud information. The tire point cloud is then converted to the test platform coordinate system based on the calibration transformation matrix T. Finally, a 3D elliptical torus model point cloud template is created based on the tire parameters. Template matching is performed on the tire point cloud using a nonlinear optimization approach. The designed objective function minimizes the sum of the 3D distances between the tire's 3D points and the template. The optimal solution is then found, ultimately determining the tire's toe angle, camber angle, and wheel center. This information is then uploaded to the control and storage terminal. The control and storage terminal synchronously receives the results from the four test devices and calculates the current vehicle body angle based on the four wheel centers. The toe angle for each wheel at this moment is equal to the toe angle minus the vehicle body angle. The acquisition and calculation frequency of the acquisition camera is 10 Hz, and the wheel speed is 2 seconds per revolution. The control and storage terminal receives and calculates 20 consecutive frames of data using the aforementioned method, storing them in a cache. The average of the 20 frames' toe angles and camber angles is calculated as the current angular posture. Combined with the tire parameters, the terminal calculates the vehicle's front wheel toe, rear wheel camber, caster, and kingpin inclination angles, and displays the resulting information on the interface. The control storage terminal calculates various parameters by updating the cache, displays them on the interface in real time, and writes the information into the database in real time for subsequent inspection. Ultimately, the present invention realizes dynamic real-time measurement with high measurement accuracy without damaging the wheel. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of the system detection of the present invention;

[0037] Figure 2 It is a calibration schematic diagram of the present invention;

[0038] Figure 3 It is a schematic diagram of the model of the present invention;

[0039] Figure 4 Schematic diagram of the point cloud of the present invention;

[0040] Figure 5 It is a schematic diagram of the overall structure of the four-wheel alignment real-time posture detection device of the present invention.

[0041] In the accompanying drawings, the components represented by the reference numerals are as follows:

[0042] 1. Test platform; 2. First slide rail; 3. First slider; 4. Roller; 5. Control storage terminal; 6. Vertical rail; 7. Second slider; 8. Test equipment; 801. Equipment body; 802. Light source; 803. Collection camera. DETAILED DESCRIPTION

[0043] In order to make the purpose and advantages of the present invention more clearly understood, the present invention is described in detail below with reference to the following examples. It should be understood that the following text is only used to describe one or more specific embodiments of the present invention and does not strictly limit the scope of protection of the present invention.

[0044] Example 1

[0045] like Figure 1-Figure 5 As shown, a vision-based four-wheel positioning real-time posture detection device includes a test platform 1 and a control storage terminal 5, wherein the test platform 1 is symmetrically provided with a first slide rail 2 on both sides, the first slide rail 2 is movably provided with a first slider 3, the first slider 3 is symmetrically and movably connected to rollers 4 at both ends, the first slider 3 is connected to a vertical rail 6 at the center position, the vertical rail 6 is movably provided with a second slider 7, the end of the second slider 7 is installed with a test device 8, and the four test devices 8 are electrically connected to the control storage terminal 5.

[0046] Preferably, distance sensors are provided at the inner end of the first slide rail 2 and the inner end of the vertical rail 6, and limiting mechanisms are provided inside the first slide rail 2 and on the inner and outer ends of the vertical rail 6. The testing equipment 8 includes an equipment body 801, a light source 802 is provided at the center position of the side of the equipment body 801, and two acquisition cameras 803 are symmetrically provided on the side of the equipment body 801. The acquisition cameras 803 are symmetrically distributed about the light source 802, and each group of two acquisition cameras 803 are respectively marked as an upper camera and a lower camera.

[0047] In this context, the primary purpose of a four-wheel alignment is to adjust the angles of the vehicle's four wheels to achieve optimal balance between the wheels, the vehicle body, and the ground, meeting design requirements. Simply put, the three-dimensional rolling attitude of the wheels is uniquely determined by two key parameters: the wheel's deviation from the vehicle's forward direction and its inclination perpendicular to the ground. These two parameters are used to calculate the vehicle's front wheel toe, rear wheel camber, castor, and kingpin inclination.

[0048] Example 2

[0049] A vision-based four-wheel alignment real-time posture detection system, the detection system is used to control the four-wheel alignment real-time posture detection device in embodiment 1, the four-wheel alignment real-time posture detection system includes a collection unit and a sensor calculation unit; the collection unit is electrically connected to the sensor calculation unit;

[0050] The acquisition unit synchronously acquires the wheel area image at the current moment through the four test devices 8 to obtain the acquired data. The sensor calculation unit performs digital processing based on the visual image according to the acquired data, and finally obtains the tire's toe direction angle, camber angle, and wheel center, and uploads the information to the control storage terminal 5.

[0051] Example 3

[0052] A vision-based four-wheel alignment real-time posture detection method, the four-wheel alignment real-time posture detection method is a method for using the detection system in Example 2 and the four-wheel alignment real-time posture detection device in Example 1; the four-wheel alignment real-time posture detection method comprises the following steps:

[0053] Step 1: Perform coordinate system calibration. Define the coordinate system of the calibration frame on the test platform 1. Each calibration plate on the calibration frame corresponds to a test device 8. Perform coordinate system transformation based on the true value and measured value of the cylindrical surface. Finally, calculate the basic transformation matrix T0 from the device coordinate system of the test device 8 to the coordinate system of the test platform 1.

[0054] Step 2: Adjust the spacing between the two sets of the first sliders 3 according to the wheelbase of the vehicle to be tested so that the wheels of the test vehicle fit on the floating plates of the two sets of rollers 4. Then adjust the height of the second slider 7 on the vertical rail 6 so that the test device 8 is aligned with the center of the wheel. Control the storage terminal 5 to send a coordinate adjustment instruction to the test device 8. The test device 8 calculates the current transformation matrix T based on T0 and loads it into the initialization cache of the calculation unit.

[0055] Step 3: The control storage terminal 5 sends instructions to the test platform 1 and the synchronizer of the sensor. The roller 4 drives the wheel to rotate. The synchronizer sends instructions to the four test devices 8. The test devices 8 start measuring. After the acquisition camera 803 collects images and calculates point cloud data, the point cloud is converted and calculated using the above matrix T. The results of the four wheels are synchronously uploaded to the control storage terminal 5.

[0056] Step 4: Control the storage terminal 5 to perform post-processing operations based on the synchronous data obtained by the test equipment 8, and calculate the vehicle's front wheel toe, rear wheel camber, kingpin caster angle, and kingpin inclination angle information in combination with the tire parameters, and upload the information to the interface and data storage library.

[0057] Preferably, in step 1, the point O between the two positioning holes N1 and N2 on the front axis of the calibration frame is defined as the origin of the coordinate system, the major axis of the calibration frame is the X axis, the minor axis is the Y axis, and the axis perpendicular to the XOY plane is the Z axis; the calibration frame is placed at the center line OK of the platform through the positioning holes of the test platform 1 by means of a pin; wherein K is the midpoint of the positioning holes N3 and N4, so that the center line is parallel to the major axis direction of the calibration frame; each calibration plate on the calibration frame corresponds to a test device 8, and the coordinate system conversion is performed according to the true value and the measured value of the cylindrical surface, wherein the true value is the three-coordinate value;

[0058] For each acquisition camera 803, the calibration plate is scanned to obtain the target area image. After the midline of the upper and lower cameras is extracted, matched, and restored, the point cloud of the calibration plate and the three cylinders A, B, and C is obtained. The plane point cloud where the three cylindrical surfaces are located is calculated based on RANSAC plane fitting. The center coordinates of the three cylinders are obtained by outlier filtering and circle fitting. A '、O B '、O C ', finally, according to the three coordinate values of the centers of the three cylinders O A , O B , O C , calculate and obtain the transformation matrix T of the test device 8 from the device coordinate system to the test platform 1 coordinate system.

[0059] In the present invention: During static measurement, due to the gravity of the vehicle itself, the deflection error caused by the force between the wheel and the ground is introduced. The current posture is not the relative posture of the wheel and the vehicle, resulting in low measurement accuracy. During dynamic measurement, if the wheel and the rotating shaft present a certain angle, the posture of the wheel will change periodically after one circle of rotation. Therefore, the positioning accuracy can be guaranteed by replacing the current wheel posture with the average value of the posture change within a week. During dynamic measurement, the posture of the vehicle itself changes in real time. The toe direction angle measured by a single acquisition camera 803 at any time includes the body posture at the current moment. How to obtain the accurate body posture is also the prerequisite for obtaining the accurate toe angle. During dynamic adjustment of the four-wheel alignment, due to the characteristic of "permanence of vision" of the human eye, the real-time synchronous result display is extremely important for the adjustment efficiency.

[0060] like Figure 2 As shown, the third test device 8 and the fourth test device 8 are installed on a guide rail parallel to the OK axis; before measurement, as shown in FIG. Figure 2As shown, during calibration, the distance between the fourth acquisition device 8 and device 2 is S0. Test platform 1 receives vehicle information, including a wheelbase of S1. The third and fourth acquisition devices 8 and their corresponding floating plates, positioned toward the rear wheels, are moved along the guideway to position S1. The guideway stops, and the vehicle is then aligned with the floating plate on test platform 1. Based on the wheel dimensions, test platform 1 calculates the rotational speed of the floating plate roller 3, achieving a rotational speed of 2 seconds per revolution. This controls the rotation of roller 4, which in turn drives the wheel, simulating a vehicle's driving state. This completes the preparations.

[0061] Preferably, the specific steps of step 4 are as follows:

[0062] Step 401: According to the instruction of the synchronizer of the control storage terminal 5, the four test devices 8 synchronously acquire the wheel area image at the current moment. The sensor computing unit performs image preprocessing based on the acquired data, extracts the sub-pixel center line of the laser line in the image, and calculates the wheel hub edge feature points;

[0063] Step 402: The wheel hub diameter and tire height are then calculated based on the uploaded tire parameters and converted to a pixel coordinate system as conditional parameters. RANSAC ellipse fitting based on edge feature points is used to locate the wheel hub edge in the image. The tire outer edge ellipse is obtained based on the tire height pixel information. The area between the two concentric ellipses is the tire tread area. The centerline outside the tread area is filtered to obtain the tire centerline information in the image.

[0064] Step 403: Based on the centerline pixel coordinates of the upper and lower cameras and the camera's extrinsic parameters, encoding-based feature point matching, filtering, and restoration operations are performed to obtain tire point cloud information. The tire point cloud is then converted to the test platform 1 coordinate system based on the calibration transformation matrix T.

[0065] Step 404: Finally, a three-dimensional elliptical ring model point cloud template is established based on the tire parameters. The tire point cloud is template matched based on a nonlinear optimization method. The sum of the three-dimensional distances between the three-dimensional points of the tire and the template is minimized through the designed objective function. The optimal solution is obtained to finally obtain the tire's toe angle, camber angle, and wheel center.

[0066] Preferably, the three-dimensional elliptical ring model constructed in step 404 is defined as follows: Figure 3 As shown, assuming that the center of a circle on the XOY plane is O r An ellipse with one axis passing through the X axis and an axis length of 2r, O r The distance from the origin O is R. The torus formed by rotating around the Z axis with O as the control point is a three-dimensional elliptical torus. Let its model formula be:

[0067]

[0068] 0≤v≤π (left) or -π≤v≤0 (right), 0≤u≤2π, -5≤k≤5, The objective function is:

[0069]

[0070] Preferably, in step 404, the point set on the tire is {x i 、y i 、z i │i=0,1,…,n}, the wheel center is (C x ,C y ,C z ), the toe angle is α, the inclination angle is γ, and the angle of rotation around the Y axis is θ. The optimization steps are:

[0071] Step 4041: Initialize the parameters to be optimized; the parameters to be optimized are C x ,C y ,C z ,α,γ,R,r,k,θ, first, according to each dense point line on the tire point cloud converted in step 403, Figure 4 Perform circle fitting on the midpoint shown to obtain the initial center point and angle C x ,C y ,C z ,α,γ, set the initial value of θ to 0, and then calculate the initial value of R,r,k according to the obtained tire size parameters;

[0072] Step 4042: Then according to C x ,C y ,C z ,α,γ calculate the rotation matrix A∈R ∧ (4×4), transform the tire point cloud coordinate system to the model coordinate system;

[0073] Step 4043: Then, the final objective function is obtained according to formula (1) and formula (2), and the optimal solution of the parameters to be optimized is obtained by constructing a nonlinear optimization objective function;

[0074] Step 4044: Finally, C x ,C y ,C z ,α,γ are uploaded to the control storage terminal 5.

[0075] Preferably, the control storage terminal 5 synchronously receives the result information of the four test devices 8, and calculates the body angle at the current moment based on the four wheel centers. At this moment, the toe angle of each wheel is equal to the toe direction angle minus the body angle; the acquisition and calculation frequency of the four test devices 8 is 10 Hz, and the wheel speed is 2s / cycle. The control storage terminal 5 receives and calculates 20 consecutive frames of data in the above-mentioned manner and puts them into the cache, calculates the average value of the 20 frames of toe angles and camber angles as the angle posture at the moment, and calculates the vehicle's front wheel toe, rear wheel camber, kingpin caster angle, and kingpin inclination angle information in combination with the tire parameters, and displays the current result information on the interface; the control storage terminal 5 calculates each parameter by updating the cache, displays it on the interface in real time, and writes the information into the database in real time for subsequent inspection.

[0076] The foregoing is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art may make various improvements and modifications without departing from the principles of the present invention, and such improvements and modifications are also within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained herein shall, unless otherwise specified or limited, be implemented in accordance with conventional means in the art.

Claims

1. A vision-based four-wheel alignment real-time posture detection device, characterized by: The invention comprises a test platform (1) and a control storage terminal (5), wherein first slide rails (2) are symmetrically provided on both sides of the test platform (1), a first slider (3) is movably provided on the first slide rail (2), rollers (4) are symmetrically and movably connected to both ends of the first slider (3), a vertical rail (6) is connected at the center of the first slider (3), a second slider (7) is movably provided on the vertical rail (6), a test device (8) is installed at the end of the second slider (7), and four test devices (8) are electrically connected to the control storage terminal (5).

2. The vision-based four-wheel alignment real-time posture detection device according to claim 1, characterized in that: The inner end of the first slide rail (2) and the inner end of the vertical rail (6) are both provided with distance sensors, and the inside of the first slide rail (2) and the inside of the vertical rail (6) are both provided with limiting mechanisms. The testing device (8) includes a device body (801), a light source (802) is provided at the center position of the side of the device body (801), and two acquisition cameras (803) are symmetrically provided on the side of the device body (801). The acquisition cameras (803) are symmetrically distributed about the light source (802), and each group of two acquisition cameras (803) are respectively marked as an upper camera and a lower camera.

3. A vision-based four-wheel alignment real-time posture detection system, characterized by: The detection system is used to control the four-wheel alignment real-time posture detection device according to any one of claims 1-2, and the four-wheel alignment real-time posture detection system includes a sensor acquisition unit and a sensor calculation unit; the sensor acquisition unit is electrically connected to the sensor calculation unit; The sensor acquisition unit synchronously acquires the wheel area image at the current moment through the four test devices (8) to obtain acquisition data. The sensor calculation unit performs digital processing based on the visual image according to the acquisition data, and finally obtains the toe direction angle, camber angle and wheel center of the tire, and uploads the information to the control storage terminal (5).

4. A vision-based four-wheel alignment real-time posture detection method, characterized by: The four-wheel alignment real-time posture detection method is a method for using the detection system according to claim 3; the four-wheel alignment real-time posture detection method comprises the following steps: Step 1: Perform coordinate system calibration, define the coordinate system of the calibration frame on the test platform (1), each calibration plate on the calibration frame corresponds to a test device (8), and perform coordinate system conversion based on the true value and measured value of the cylindrical surface; finally, calculate the basic conversion matrix T0 of the test device (8) from the device coordinate system to the test platform (1) coordinate system; Step 2: adjusting the distance between the two groups of the first sliders (3) according to the wheelbase of the vehicle to be tested so that the wheels of the tested vehicle fit on the floating plates of the two groups of rollers (4), then adjusting the height of the second slider (7) on the vertical rail (6) so that the test device (8) is aligned with the center of the wheel, controlling the storage terminal (5) to send a coordinate adjustment instruction to the test device (8), and the test device (8) calculates the current conversion matrix T in combination with T0; Step 3: The control storage terminal (5) sends instructions to the test platform (1) and the synchronizer of the sensor, the roller (4) drives the wheel to rotate, and the synchronizer sends instructions to the four test devices (8). The test devices (8) start measuring, and after the image acquisition camera (803) is used to collect and calculate the point cloud data, the coordinate system is converted by the above matrix T, and then the toe direction angle, camber angle and wheel center of each wheel are calculated, and the results are synchronously uploaded to the control storage terminal (5); Step 4: Control the storage terminal (5) to perform post-processing operations based on the synchronous data obtained by the test equipment (8), calculate the vehicle's front wheel toe, rear wheel camber, kingpin caster angle, and kingpin inclination angle information in combination with tire parameters, and upload the information to the interface and database storage.

5. The method for real-time posture detection of four-wheel alignment based on vision according to claim 4, characterized in that: The specific steps of step 4 are as follows: Step 401: According to the instruction of the synchronizer of the control storage terminal (5), the four test devices (8) synchronously acquire the wheel area image at the current moment, and the sensor calculation unit performs image preprocessing based on the acquired data, extracts the sub-pixel center line of the laser line in the image, and calculates the wheel hub edge feature points; Step 402: The wheel hub diameter and tire height are then calculated based on the uploaded tire parameters and converted to a pixel coordinate system as conditional parameters. RANSAC ellipse fitting based on edge feature points is used to locate the wheel hub edge in the image. The tire outer edge ellipse is obtained based on the tire height pixel information. The area between the two concentric ellipses is the tire tread area. The centerline outside the tread area is filtered to obtain the tire centerline information in the image. Step 403: Based on the centerline pixel coordinates of the upper and lower cameras and the external parameters of the cameras, encoding-based feature point matching, filtering, and restoration operations are performed to obtain the tire point cloud information, and the tire point cloud is converted to the test platform (1) coordinate system according to the calibration transformation matrix T; Step 404: Finally, a three-dimensional elliptical ring model point cloud template is established based on the tire parameters. The tire point cloud is template matched based on a nonlinear optimization method. The sum of the three-dimensional distances between the three-dimensional points of the tire and the template is minimized through the designed objective function. The optimal solution is obtained to finally obtain the tire's toe angle, camber angle, and wheel center.

6. The method for real-time posture detection of four-wheel alignment based on vision according to claim 5, characterized in that: The three-dimensional elliptical ring model constructed in step 404 is defined as follows: Let the center of a circle on the XOY plane be O r An ellipse with one axis passing through the X axis and an axis length of 2r, O r The distance from the origin O is R. The torus formed by rotating around the Z axis with O as the control point is a three-dimensional elliptical torus. Let its model formula be: 0≤v≤π (left) or -π≤v≤0 (right), 0≤u≤2π, -5≤k≤5, The objective function is:

7. The method for real-time posture detection of four-wheel alignment based on vision according to claim 6, characterized in that: In step 404, the point set on the tire is {x i 、y i 、z i │i=0,1,…,n}, the wheel center is (C x ,C y ,C z ), the toe angle is α, the inclination angle is γ, and the angle of rotation around the Y axis is θ. The optimization steps are: Step 4041: Initialize the parameters to be optimized; the parameters to be optimized are C x ,C y ,C z ,α,γ,R,r,k,θ, firstly perform circle fitting based on the midpoint of each dense point line on the tire point cloud converted in step 403 to obtain the initial center point and angle C x ,C y ,C z ,α,γ, set the initial value of θ to 0, and then calculate the initial value of R,r,k according to the obtained tire size parameters; Step 4042: Then according to C x ,C y ,C z ,α,γ calculate the rotation matrix A∈R ∧ (4×4), transform the tire point cloud coordinate system to the model coordinate system; Step 4043: Then, the final objective function is obtained according to formula (1) and formula (2), and the optimal solution of the parameters to be optimized is obtained by constructing a nonlinear optimization objective function; Step 4044: Finally, C x ,C y ,C z ,α,γ are uploaded to the control storage terminal (5).

8. The method for real-time posture detection of four-wheel alignment based on vision according to claim 5, characterized in that: The control storage terminal (5) synchronously receives the result information of the four test devices (8), and calculates the body angle at the current moment based on the four wheel centers. At this moment, the toe angle of each wheel is equal to the toe direction angle minus the body angle; the acquisition and calculation frequency of the four test devices (8) is 10 Hz, and the rotation speed of the wheel is 2 s / cycle. The control storage terminal (5) receives and calculates 20 consecutive frames of data in the above-mentioned manner and puts them into the cache, calculates the average value of the 20 frames of toe angles and camber angles as the angle posture at the moment, and calculates the front wheel toe, rear wheel camber angle, kingpin caster angle, and kingpin inclination angle information of the vehicle in combination with tire parameters, and displays the result information on the interface; the control storage terminal (5) calculates each parameter by updating the cache, displays it on the interface in real time, and writes the information into the database in real time for subsequent inspection.

Images