HK1242412B - Rolling virtual wheel spindle calibration - Google Patents

Description

Virtual Axle Calibration in Rolling

Technical Field

Embodiments generally relate to systems and methods for vehicle wheel alignment. The subject matter has particular application in determining true values of wheel alignment parameters, such as camber and toe angle, for a vehicle when using an image aligner having a target attached to a vehicle wheel and a camera that images the target.

Background Art

Machine vision vehicle alignment systems (also known as "image aligners") that use a movable camera and a target attached to a vehicle wheel are well known. The target is viewed by a camera, and the image data obtained for a prescribed alignment process can be used to calculate vehicle alignment angles for display via a user interface (typically a computer monitor). Early system implementations included a rigid beam connecting the cameras so that their position and orientation relative to each other could be determined and, since they were constant, relied upon. Later system implementations included cameras that were not rigidly connected to each other, but used independent camera/target systems to continuously calibrate the position of one vehicle-mounted target observation camera relative to the other. This type of system is described in U.S. Patents 5,535,522, 6,931,340, 6,959,253, and 6,968,282, the entire contents of which are incorporated herein by reference. An example of a vehicle wheel aligner that uses this type of image processing is the Visualiner 3D, or "V3D," commercially available from John Bean Company (a division of Snap-on Incorporated) in Conway, Arkansas.

In order to be able to use an image aligner to accurately measure the wheel alignment angle for a vehicle, the wheel's axis of rotation about which the target rotates must be measured, and the coordinates of the virtual wheel axle point through which the vector passes must be determined.

A conventional method for calibrating a combined target and fixture system involves lifting the vehicle off a supporting surface (e.g., a workshop floor or a gantry) so that the wheel on which the target is mounted can rotate freely. The wheel is then rotated to a predetermined position so that the vector defining the wheel's axis of rotation can be determined. Because the target origin intersects a circular arc, the coordinates of the circle's center are calculated based on points on the arc's circumference. The center of the circle lies on the wheel's axis of rotation and is called the virtual axle point. The virtual axle point is projected along the axle axis onto the plane of the wheel rim. The projected point is the axle point. This is illustrated in Figure 1, where a fixture 110 carrying a target 120 is mounted on the wheel of a vehicle 100. The coordinates are in the target coordinate system. The axis of rotation vector 130 passes through the axle point 140 and the virtual axle point 150. The target's center of mass is offset from the virtual axle point 150.

Typical wheel fixtures and target assemblies are manufactured so that the calibration process does not have to be repeated each time the fixture is removed from the wheel. For this reason, conventional target assemblies typically use self-centering wheel fixtures. Fixture installation errors are compensated for by well-known roll and run calculations.

When the locator is first set up, a calibration procedure for the system of targets and self-centering fixtures is typically performed by a technician using custom calibration equipment. This calibration procedure must then be performed each time a new target is introduced into the system (e.g., when the target is replaced). Disadvantageously, the end user must wait for a service technician to arrive, or if the end user wants to perform the calibration procedure themselves, they must undergo special training. Furthermore, during normal use, the target and its associated fixture are susceptible to changing their relative geometric relationship (e.g., if the fixture is dropped). Although the fixture and target may still be usable, this change in relative geometry is not reflected in the original system calibration, and thus disadvantageously results in a decrease in positioning accuracy over time.

There is a need for a method and apparatus that can determine the wheel axis of rotation and axle point without requiring the additional time required to lift the vehicle and perform additional procedures not necessary for typical alignment. There is also a need for a method and apparatus that can accommodate normal wear of the wheel target assembly to maintain alignment accuracy. Furthermore, there is a need to minimize the cost of the wheel fixture by eliminating the need for self-centering capabilities.

Summary of the Invention

The disclosed system and method determine the wheel's axis of rotation and axle point each time an alignment is performed by rolling the vehicle wheel and tracking the motion of a target attached to the wheel. The disclosed procedure can be performed concurrently with the conventional roll and yaw compensation routine (which is part of the standard wheel alignment process flow) and must in any case be performed near the beginning of the alignment procedure. Furthermore, the present invention is able to calculate the axle point and axis of rotation regardless of the radial location of the target on the wheel, thereby eliminating the need for a self-centering wheel fixture.

More specifically, the present invention discusses determining wheel alignment angles by measuring the attitude of a target as the wheel rolls without slip. A camera system tracks a target positioned at a radius between the wheel axle and the wheel circumference. Specifically, the target's origin and orientation are tracked. In the ideal case of non-slip 2-D motion, as the wheel rolls, the trajectory of the target's origin follows a path known as a cycloid. The wheel's rotational angle is determined based on the target's changing attitude.

The virtual axle point can be calculated based on the motion of the target origin as the wheel rotates. When the coordinates and pose of the three targets are known without measurement error, calculating the motion of the virtual axle point has an analytical solution. The equation for the target's path can be determined by measuring the target coordinates and pose, and a model can be used to fit the data to determine the parameters of the equation. The path of the virtual axle point can then be calculated based on the parameters of the equation.

One or more embodiments include a wheel alignment method for a vehicle, the method comprising: attaching a target to a wheel of the vehicle, and providing a camera for viewing the target and capturing image data of the target. The vehicle is driven while the wheel is on a substantially flat surface to rotate the wheel and the target by a plurality of degrees while the camera captures image data of the target. A wheel rotation axis and a wheel axle point are calculated based at least in part on the captured image data. Alignment parameters for the vehicle are calculated using the wheel axle point and the wheel rotation axis.

Embodiments also include a vehicle wheel alignment system comprising: a target fixedly attachable to a wheel of a vehicle; a camera for viewing the target and capturing image data of the target; and a data processor. The data processor is adapted to receive the image data from the camera and determine a vector pointing from a target origin to a wheel axle point based at least in part on the image data of the target captured while the vehicle is driven with the wheel on a substantially flat surface, thereby rotating the wheel and the target by a plurality of degrees. The data processor is further adapted to calculate alignment parameters for the vehicle based at least in part on the coordinates of the wheel rotation axis and the wheel axle point.

An embodiment also includes a non-transitory computer-readable medium having stored thereon instructions that, when executed by a processor of a vehicle positioning system, cause the processor to determine positioning parameters for the vehicle. The positioning system includes a target that can be fixedly attached to a wheel of the vehicle; and a camera for viewing the target and capturing image data of the target. The determination process includes receiving image data from the camera; determining coordinates of a wheel rotation axis and a wheel axle point based at least in part on image data of the target captured while the vehicle is driven with the wheel on a substantially flat surface, thereby rotating the wheel and the target by a plurality of degrees; and calculating the positioning parameters for the vehicle based at least in part on the coordinates of the wheel rotation axis and the wheel axle point.

Objects and advantages of embodiments of the disclosed subject matter will become apparent from the following description when considered in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Hereinafter, the embodiments will be described in detail with reference to the accompanying drawings, wherein like reference numerals represent like elements. The accompanying drawings are not necessarily drawn to scale. Where applicable, some features may not be shown in order to facilitate description of the underlying features.

1 is a perspective view of a vehicle with a fixture/target assembly attached showing the virtual axle point, wheel axle point, and wheel rotation axis.

FIG. 2 is a schematic diagram illustrating an example of how to calculate the coordinates of a virtual wheel axle point located in a motion plane of a target origin when the wheel is freely rotating according to various embodiments.

FIG. 3 is a schematic diagram illustrating a path of a target when a wheel to which the target is attached rolls.

4 is a schematic diagram illustrating geometric parameters of a rolling wheel with a target attached, including parameters used in an exemplary derivation of initial virtual wheel axle point coordinates, according to various embodiments.

5A-5B are graphs illustrating the rotation of 2-D target pose data into the Y-Z plane during pre-processing according to various embodiments.

6A-6D are graphs illustrating fitting iterations when adjusting parameters in a nonlinear least squares fitting calculation according to various embodiments.

FIG. 7 is a graph illustrating measured coordinates of a target origin and a virtual axle point according to various embodiments.

FIG. 8 is a graph illustrating improved consistency between virtual axle point coordinates and virtual axle point coordinates determined by fitting coordinate data according to various embodiments.

9A-9B are graphs illustrating outlier data points detected after fitting a short cycloid to measured data, according to various embodiments.

10 is a schematic top plan view of a self-calibrating wheel alignment system capable of implementing the disclosed systems and methods.

11 is a schematic top plan view of a hybrid wheel alignment system capable of implementing the disclosed systems and methods.

DETAILED DESCRIPTION

It should be understood that the principles described herein are not limited in application to the details of construction or the arrangement of components set forth in the following description or illustrated in the following figures. These principles may be embodied in other embodiments and may be implemented or carried out in various ways. Furthermore, it should be understood that the phraseology and terminology used herein are for descriptive purposes only and should not be considered as limiting.

Disclosed herein are methods and systems for wheel axis vector calculation. FIG10 is a schematic top plan view of certain elements of a computer-aided 3D (three-dimensional) motor vehicle wheel alignment system (“aligner”), such as the alignment system disclosed in U.S. Patent 6,968,282 discussed above. Such an alignmenter has the same elements as the alignmenter disclosed herein and can be used to implement the disclosed technology. In particular, the alignmenter of FIG10 includes a left camera module 2 and a right camera module 4 for locating the wheels of a motor vehicle. The terms “left” and “right” are used for convenience and are not intended to require that particular elements be positioned in a particular position or relationship relative to other elements.

Arrow 30 in Figure 10 schematically represents the motor vehicle being positioned. The vehicle includes a left front wheel 22L, a right front wheel 22R, and a left rear wheel 24L, a right rear wheel 24R. Positioning targets 80a, 80b, 80c, and 80d are secured to each of the wheels 22L, 24L, 22R, and 24R, respectively. Each positioning target typically includes a plate 82 imprinted with target information and a clamping mechanism 88 for securing the target to the wheel. The left camera module 2 includes a left positioning camera 10L. The left positioning camera 10L faces the vehicle and observes the left targets 80a and 80b along axis 42. The right camera module 4 includes a right camera 10R, which faces the vehicle and observes the right targets 80c and 80d along axis 44. The left camera module 2 also includes a calibration camera 20, which is mounted perpendicular to the camera 10L using a bracket 12. Calibration camera 20 views calibration target 16 attached to right camera module 4 by bracket 14 along axis 46 to determine the position of positioning cameras 10L, 10R relative to each other. Cameras 10L, 10R, 20 have illumination sources 62, 64, 66, respectively.

The disclosed locator also includes a data processor (not shown), such as a conventional personal computer (PC), having software with instructions that cause the data processor to electronically perform the calculations described herein.

The methods and apparatus described herein are also suitable for use with the hybrid locator system described in US Pat. No. 7,313,869, which is hereby incorporated by reference in its entirety. FIG11 shows a schematic diagram of an exemplary hybrid locator system capable of use with the methods and apparatus disclosed herein. The locator system includes a pair of passive targets 21 and 23 mounted on respective wheels 22 and 24 of a vehicle (in this first embodiment, the front steering wheels). A pair of active sensing heads 25 and 27 are suitable for mounting in association with the vehicle's other respective wheels 26 and 28 (in this case, the rear wheels). Each active sensing head includes a camera 29 or 31 for generating 2D image data; when each sensing head is mounted to a respective wheel of the vehicle, the 2D image data is expected to include an image of one of the targets 21 and 23. The system also uses two conventional (1D) angle sensors 33 and 35 to measure the relative angle of the active sensing heads 25 and 27 in the toe plane, and a pair of tilt sensors 37 and 39 to measure the tilt (typically camber and pitch) of the active sensing heads 25 and 27.

definition

Target coordinate system: A coordinate system defined by the geometric parameters of the target.

Target Origin: The mathematical point defined as the origin of the target coordinate system.

Wheel Rotation Axis: The axis about which the wheel rotates. Also known as the axle axis.

Rim Plane: The plane defined by the outer surface of the rim.

Virtual Axle Point: The point along the wheel's axis of rotation around which the target origin rotates.

Axle Point: The point where the wheel's axis of rotation intersects the plane of the rim.

Target radius: The distance between the target origin and the virtual wheel axle point.

Camera tilt angle: The tilt angle of the camera relative to the direction of motion of the wheel axle.

Initial rolling angle: The angular position of the target origin on the wheel when rolling begins.

Jacked Axle Alignment: The process or result of calculating the wheel's axis of rotation and virtual axle point based on measurements made when the wheel is raised so that it can rotate freely without moving in a straight line.

Rolling Axle Alignment: The process or result of calculating the wheel's axis of rotation and virtual axle point based on measurements made while the wheel is rolling.

The disclosed rolling axle calibration technique offers several significant advantages over conventional jacked-up axle calibration procedures. One advantage is that rolling axle calibration is an "online" process rather than an "offline" process. Rolling axle calibration is performed every time roll runout compensation is performed and is part of the standard wheel alignment process flow. Unlike jacked-up axle calibration, there are no special procedures that the end user must follow and no special training is required when having to perform this system calibration. As a result, highly trained field service personnel with customized calibration equipment are not required to perform high-precision system calibrations. This saves the end user time and money.

Another advantage of being an online process is that the actual axle alignment changes over time. During normal use, the target and fixture can easily change their relative geometry (for example, if the fixture is accidentally dropped). This change in relative geometry is not a problem; what is important is that the relative geometry is accurately reflected in the axle alignment. As an online measurement, the rolling axle alignment provides up-to-date measurements of the axle point coordinates and the orientation of the wheel's axis of rotation.

Another advantage of rolling axle alignment is felt when a vehicle-centric wheel alignment coordinate system is employed. In a conventional vehicle-centric coordinate system, the coordinate system is constructed based on measurements of the vehicle being inspected (typically using the center of the vehicle's rim). When employing the disclosed rolling axle alignment, no fixed spatial relationship between the target and the rim center is assumed. The center of the wheel rim is calculated as part of the process. Therefore, the target can be placed at any relative radial position with respect to the wheel axis. In other words, the target can be attached to the wheel so that the target origin is approximately arranged on the wheel's axis of rotation, or so that the target origin is offset from the wheel's axis of rotation. Therefore, the disclosed rolling axle alignment technique provides a certain degree of freedom in the type of target used.

Overview

The purpose of axle alignment measurements is to determine the location of the axle point relative to the target for all positions and orientations of the wheel to which the target is attached. The axle point is the point located at the intersection of the wheel's axis of rotation and the plane defined by the outward-facing surface of the wheel rim. The origin of the target coordinate system is typically referred to as the target origin. Those skilled in the art will appreciate that determining the location of the axle point and the orientation of the wheel's axis of rotation are integral to determining wheel and frame alignment.

The measurable quantity for any wheel position is the orientation of the target origin and its attitude. The wheel axis of rotation is calculated based on the change in orientation of the target attitude as the wheel rotates. The coordinates of the wheel axle point in the target coordinate system remain unchanged as the wheel rotates.

The conventional procedure for determining the axle point and wheel rotation axis is to lift the vehicle until the wheels can rotate freely. Each wheel is then rotated to at least two positions while observing a target attached to the wheel and measuring the target pose. Measuring the target pose at these two positions, as well as the coordinates of the target origin, allows calculation of the rotation angle between the target poses, the vector defining the wheel rotation axis, the target radius, and the virtual axle point. The geometric relationships are shown in Figure 2, where:

X: 1/2 of the chord length between the measurement points

θ: rotation angle of the target

P ₁ , P ₂ : Coordinates of the measured target center of mass

C: The center of the circle with coordinates (x ₀ , y ₀ )

When the target origin is rotated about the wheel's axis of rotation, the plane defined by the target origin becomes parallel to the plane defined by the face of the wheel rim (in which the axle point lies). The distance between these two planes is called the target offset distance and is determined by the fixture's geometry. This distance is used to calculate the coordinates of the axle point relative to the target origin.

Raising the vehicle, rotating the wheel while taking measurements, and lowering the vehicle are error-prone, time-consuming, labor-intensive, and expensive processes that users want to avoid. Using the conventional procedure just described, the vehicle needs to be raised each time the combined system of a target and its fixture needs to be calibrated. Systems using self-centering fixtures are used in conjunction with the target and calibrated once for subsequent vehicles. An important feature of self-centering fixtures is that they can be placed on the wheel so that the target's translation relative to the virtual wheel axle point is fixed. However, self-centering fixtures impose undesirable size, appearance, and cost constraints.

The cost of lifting the vehicle and the cost of the self-centering fixture have led to the desire to develop alternative methods to perform wheel axle calibration without lifting the vehicle. It would be advantageous to calculate the wheel axle point and the wheel rotation axis while the vehicle wheels are rolling on the ground without slipping. This is already part of the process used to calculate the wheel alignment (e.g., conventional roll and yaw compensation), and no special calibration steps are required in the process. An advantage of the disclosed wheel axle calibration while rolling is that the fixture/target system does not need to have fixed self-centering geometric parameters. The requirement to calibrate such a system at each use means that only a simple calibration procedure is required.

Calibrating the raised wheels

When the wheel is raised and rotated, the trajectory of the target located at a radial distance from the wheel's axis of rotation (the virtual axle point) is circular. The radius and position of the virtual axle point can be calculated from the two target coordinates and the rotation angle or center angle between them. This is shown in Figure 2 below:

1. The wheel rotation axis is perpendicular to the plane containing the two measuring points P1 and P2 and the rotation center.

2. The wheel rotation axis is located on the perpendicular bisector b of the chord between the two measuring points P1 and P2.

3. The chord length (2x) is known.

4. The perpendicular bisector b of the chord bisects the known angle θ between the two measurement points P1 and P2. First, solve for the radius R based on θ and x. Next, find the intersection of the two circles with radius R, centered at P1 and P2. Although there are two solutions for the circle center, only one falls on the correct side of the line between P1 and P2.

Axle alignment during rolling

The wheel axle calibration in rolling according to the present invention is now described. When the wheel rolls without slipping in contact with the ground, the target trajectory is a short cycloid. As shown in Figure 3, the path 300 of the target origin t represents a short cycloid. For practical reasons, only a small part of the path 300 is measured, such as a 90-degree range of motion 310. The reason why the measurement range 310 is limited is that the camera 320 always needs to observe the face of the target (not shown, but can be similar to the target 120). The center of rotation of the wheel 330 and its direction of movement are not measured features. The measurement of the position and orientation of the target uses a single-viewpoint n-point attitude evaluation. At each target position, the coordinates and orientation of the target are calculated. The direction of the wheel rotation axis around which the rotation occurs is calculated based on the change in attitude.

The measurement in the 2D plane consists of 3 coordinate measurements and a pose measurement at each position. From these measurements the following parameters are calculated:

1. The diameter of the wheel 330;

2. Target radius;

3. The starting position (X, Y, Z coordinates) of the wheel 330 relative to the camera 320;

4. The rotation angle of the target attached to the wheel 330 when the wheel is in its starting position;

5. The tilt angle A of the camera or the direction of travel of the wheel 330 relative to the horizontal line defined by the camera axis.

The parameters of the short cycloid in the plane can be accurately solved from the three measurement points and the angular differences between them. Figure 4 shows the geometric parameters of a rolling wheel with a target attached at radius r _t and describes the derivation of the initial wheel axle coordinates, where:

r _t : target radius

R _ON : The rotation matrix that will be rotated to

r _ω θ _ON : linear travel of the wheel

A vector indicating the direction of linear travel

I: Identity matrix

A dash above a variable indicates that the variable is a vector.

The coordinates of the initial axle point are calculated as a function of the coordinates of the target origin and the rotation angle. Given the initial axle point coordinates, the direction of travel, and the rotation angle, the axle point coordinates for each position of the target can also be calculated.

It should be noted that the disclosed rolling axle calibration technique can also handle the case of an elevated wheel. The arbitrary linear travel of the wheel is defined by _the term _rωθON . When this term is set to zero in the implementation software, all data points lie on the circumference. The software then proceeds to calculate the fixed axle point _WO and the target radius.

Data preprocessing

The measured data points are sorted in order of increasing Z coordinate. Therefore, in the disclosed technique, whether the wheel rolls forward or backward is irrelevant. This sorting defines the order of the measured angular differences in varying target poses. With respect to the 3-point (three-point) formula, the initial wheel axle point coordinate W ₀ is closest to the camera. The simulated data generated during the nonlinear least-squares search described below uses the angular differences extracted from the measured data.

The estimation of the rolling parameters of the wheel is essentially a two-dimensional problem, with small deviations of the target origin's motion from the two-dimensional plane due to noise, and small spiral motions due to the presence of vehicle toe. The wheel's trajectory can be considered to have roll, pitch, and yaw relative to the camera. The roll and yaw parameters can be calculated in advance, while the pitch is retained as an unknown in the data. Pitch corresponds to any top-down view of the camera relative to the linear motion of the wheel and is referred to herein as the camera tilt angle (indicated by the reference symbol A in Figure 3). Roll corresponds to camera rotation, and yaw corresponds to any left-right orientation of the camera relative to the linear motion of the wheel.

From one pose to another, roll and yaw can be determined based on the measured rotation axis of the target. Figures 5A and 5B show the data rotated into the Y-Z plane during preprocessing. The Y-Z plane is determined by the camera coordinate system. The data in Figure 5A is translated and rotated into the Y-Z plane for processing in Figure 5B. For three-point fitting and nonlinear fitting, preprocessing eliminates some of the computational burden of the problem.

Nonlinear least squares search for parameters

For practical reasons, more than 3 attitude measurements may be required to perform the disclosed wheel axle calibration in rolling. It will be appreciated by the skilled person that measured noise may affect image processing, the vehicle may not move in a straight line (the wheels may turn), the platform on which the vehicle rides may include protrusions, and/or the platform on which the vehicle rides may slip so that the wheels may not experience pure rotation. Alternatively, the wheels may be slightly jerky due to gaps between the plate and the platform. Additionally, the range of motion of the vehicle may be limited by mechanical constraints. One consequence of these complications is that 3 attitude solutions may be prone to error. Therefore, the disclosed technique includes more data points and fits a parameterized curve of a short cycloid to the measured data. In this way, the data can be processed to detect and compensate for unexpected motion and other complications.

According to certain embodiments, if more than 3 measurement points are acquired during the vehicle's travel, the data are processed using a known least squares fitting method (the solution is exact even with only 3 points). The following situations may adversely affect the results:

1. As the wheel rolls from one position to the next, it may turn slightly. If there are more than 3 points, the measurement points may not be in the same plane. In addition, the direction of the wheel's axis of rotation will change.

2. As the wheels roll from one position to the next, the vehicle may encounter bumps or slide without rolling.

3. The wheels may not be perfectly round, and it is unclear how wheel deformation affects the target's motion.

4. Tire tread may cause changes in target motion.

5. There may be errors in the target's attitude angle measurement, which will cause errors in the estimated wheel rotation angle.

For these reasons, the present invention implements a numerical optimization method to estimate wheel parameters and axle position during rolling. The numerical optimization method minimizes the error between measured and simulated data by adjusting model parameters. The model parameters include the wheel diameter, target radius, the direction of the wheel's linear motion relative to the camera axis, and the wheel's starting position relative to the camera. The model parameters are then used to calculate the position of the axle point during rolling. The measured pose coordinates and orientation are then used to calculate the axle point and the wheel's rotation axis.

In some embodiments, the model parameters are determined using the well-known Nelder-Mead optimization method by minimizing the total RMS error, defined as the difference between the measured target coordinates and the simulated model coordinates. This method performs a nonlinear least squares fit of the simulated data to the measured data. Figures 6A to 6D illustrate four iterations of the fitting process as the error is minimized, with the measured data represented by reference numeral 600 and the coordinates generated by the fitting function represented by reference numerals 610 to 640. Parameters that can be adjusted include wheel diameter, target radius, the closest position of the wheel during vehicle travel, the orientation of the target at its closest position, and the orientation of the camera relative to the direction of travel.

The parameters that are varied during the fitting process are:

1. Wheel axle Y coordinate.

2. Wheel axle Z coordinate.

3. The direction of wheel movement relative to the camera axis (ie, camera tilt angle).

4. The angular position of the target when rolling begins.

5. The radius of the target origin relative to the wheel's axis of rotation.

6. Wheel diameter.

The fitting of the short cycloid to the measured data is sensitive to noise. When the added noise is random, the fitting algorithm always converges and provides the correct wheel axle coordinates, wheel diameter, and radius for the simulated data. Although the parameters are adjusted to minimize the overall error using real data, the result may contain errors in the parameters. The error will manifest as an offset in the position and orientation of the wheel axle coordinates relative to the target coordinates. This offset is presented in FIG7 as the difference between the virtual wheel axle point coordinates 700 calculated by the fitting and the virtual wheel axle point coordinates 710 determined by the calibrated target/fixture system (i.e., based on fixture calibration during rolling).

FIG8 illustrates the improved agreement between a virtual axle point 800 determined by fitting the coordinate data and a reference virtual axle point 810 determined when using a self-centering fixture. In this embodiment, higher-order terms are added to the model to account for wheel motion that is not entirely proportional to the wheel's rotational motion. These additional terms improve fitting errors and the errors in the axle point coordinates. In an unconstrained environment with a sliding and rotating platform, fitting the target origin coordinates to the model may not be robust enough.

One approach is to remove degrees of freedom from the measurement. One approach is to use a fixture (although not self-centering) to position the target origin close to the wheel's axis of rotation.

Another approach is to augment the target measurement with another measurement that defines the direction of travel of the wheel's axis of rotation. An example is a visible target pattern on the floor, which provides information about the floor's posture. Other examples include markers attached to the vehicle as it moves, or signals from an electronic level attached to the target. This additional measurement eliminates the camera tilt angle as an unknown parameter.

Alternative method for calculating wheel axle alignment in rolling

To calculate the wheel axle alignment during rolling, it is not necessary to use the Nelder-Mead simplex algorithm described in the above embodiments. Those skilled in the art will appreciate that analytical parameter models, including gradient descent, Levenberg-Marquardt, and other nonlinear least squares iterative methods, can be used to calculate the wheel axle alignment during rolling. The class of algorithms collectively referred to as "grid search" algorithms constitutes another group of available alternatives. Grid search algorithms are non-parametric parameter estimation algorithms that are typically employed when an optimization process cannot be guaranteed to occur in a purely convex space. Several alternative methods may be used in other embodiments instead of the Nelder-Mead simplex algorithm.

Outlier removal

The cycloid trajectory outlined by the target origin is a smooth arc. Bumps in the wheel motion may be detected as outliers in the data. In certain embodiments shown in Figures 9A to 9B, outlier detection is achieved by performing an initial fit of the cycloid on the data (Figure 9A). If the total error is large, it is decided to remove the data point 910 with the largest error and repeat the process until the error is small (see fit 920 in Figure 9B). This process of generating a fit on the data and removing outliers can be repeatedly run until the total error is below a threshold and there are enough remaining points.

As previously noted, the standard-required roll compensation routine can be performed concurrently with the disclosed wheel axle calibration, determining the wheel axle point coordinates and the wheel rotational axis direction vector based, at least in part, on image data captured while the vehicle is traveling. Exemplary techniques for determining roll yaw are described in U.S. Patent No. 5,535,522, discussed above, at column 12, lines 5 to 30. Those skilled in the art will appreciate that other conventional roll yaw techniques may be employed. The wheel axle point coordinates, the wheel rotational axis direction vector, and the roll yaw calculations can all be used to calculate vehicle alignment parameters, such as toe, camber, etc., in a conventional manner.

Embodiments of the method, system, and computer program product for wheel axle determination during rolling may be implemented on a general purpose computer, a special purpose computer, a programmed microprocessor or microcontroller and peripheral integrated circuit components, an ASIC or other integrated circuit, a digital signal processor, hard-wired electronic or logic circuits (e.g., discrete component circuits), a programmed logic device (e.g., a PLD, PLA, FPGA, PAL), etc. Generally, any process capable of performing the functions or steps described herein may be used to implement embodiments of the method, system, or computer program product for wheel axle determination during rolling.

Furthermore, embodiments of the disclosed method, system, and computer program product for determining a wheel axle during rolling can be readily implemented, in whole or in part, in software using, for example, an object-oriented or object-oriented software development environment that provides portable source code that can be used on a variety of computer platforms. Alternatively, embodiments of the disclosed method, system, and computer program product for determining a wheel axle during rolling can be implemented, in part or in whole, in hardware using, for example, standard logic circuits or a VLSI design. Embodiments can be implemented using other hardware or software, depending on the speed and/or efficiency requirements of each system, the specific functionality, and/or the specific software or hardware systems, microprocessors, or microcomputer systems used. Embodiments of the method, system, and computer program product for determining a wheel axle during rolling can be implemented in hardware and/or software by a person of ordinary skill in the art, having a general knowledge of computer and/or wheel alignment technology, and using any known or later developed system or structure, device, and/or software in the field to which the functional description provided herein applies.

It is therefore apparent that methods, systems, and computer program products for performing wheel axle determination during rolling are provided in accordance with the present invention. While the present invention has been described in conjunction with a number of embodiments, it is apparent that many alternatives, modifications, and variations will be or will be apparent to those skilled in the art. Applicants therefore intend to encompass all such alternatives, modifications, equivalents, and variations that fall within the spirit and scope of the present invention.

Claims (25)

1. A wheel alignment method for a vehicle, the method comprising: The target is attached to the wheel of the vehicle; A camera is provided for observing the target and capturing image data of the target; The vehicle is driven to rotate the wheels and the target, while the camera captures image data of the target. The wheel rotation axis is calculated at least in part based on the captured image data; A virtual wheel axis point in the motion plane of the target origin is calculated based at least in part on the captured image data, the target origin rotating around the virtual wheel axis point; The virtual axle point and wheel rotation axis are used to calculate the positioning parameters for the vehicle. 2. The method of claim 1, further comprising: calculating axle points based at least in part on captured image data, and using the axle points to calculate positioning parameters for the vehicle. 3. The method of claim 1, further comprising: calculating the rolling yaw of the wheel based at least in part on captured image data, and using the rolling yaw calculation to calculate positioning parameters for the vehicle. 4. The method according to claim 1, comprising: using a nonlinear least squares iterative method to calculate the virtual wheel axle point. 5. The method according to claim 4, wherein the nonlinear least squares iterative method includes one of the Nelder-Mead simplex algorithm, the Levenberg-Marquardt algorithm, and the gradient descent algorithm. 6. The method according to claim 1, comprising: using a grid search algorithm to calculate the virtual wheel axle point. 7. The method according to claim 1, comprising: attaching the target to the wheel such that the origin of the target is offset from the axis of rotation of the wheel. 8. The method according to claim 1, comprising: attaching the target to the wheel such that the origin of the target is approximately arranged on the axis of rotation of the wheel. 9. A vehicle wheel alignment system, comprising: A target that can be securely attached to the wheels of a vehicle; A camera for observing the target and capturing image data of the target; and Data processors, which are suitable for: Receive the image data from the camera; The wheel rotation axis is determined at least in part based on image data of the target captured as the vehicle moves, causing the wheels and the target to rotate. The virtual axle point is determined at least in part based on image data of the target captured while the vehicle is in motion; and The positioning parameters for the vehicle are calculated at least in part based on the wheel rotation axis and the virtual wheel axle point. 10. The system of claim 9, further comprising: calculating axle points based at least in part on captured image data, and using the axle points to calculate positioning parameters for the vehicle. 11. The system of claim 9, wherein the data processor is adapted to: calculate the rolling yaw of the wheel based at least in part on the captured image data, and calculate positioning parameters for the vehicle based at least in part on the rolling yaw. 12. The system of claim 9, wherein the data processor is adapted to use a nonlinear least squares iterative method to calculate the virtual wheel axle point. 13. The system according to claim 12, wherein the nonlinear least squares iterative method includes one of the Nelder-Mead simplex algorithm, the Levenberg-Marquardt algorithm, and the gradient descent algorithm. 14. The system of claim 9, wherein the data processor is adapted to use a grid search algorithm to calculate the coordinates of the virtual wheel axle point. 15. The system of claim 9, comprising: a clamp for attaching the target to the wheel such that the origin of the target is offset from the axis of rotation of the wheel. 16. The system of claim 9, further comprising: a clamp for attaching the target to the wheel such that the origin of the target is substantially arranged on the axis of rotation of the wheel. 17. The system of claim 9, wherein the data processor is adapted to: compare the calculated virtual wheel axle point coordinates with predetermined reference virtual wheel axle point coordinates, and notify the user when the calculated virtual wheel axle point coordinates exceed the range of the reference virtual wheel axle point coordinates. 18. A non-transitory computer-readable medium having instructions stored thereon, the instructions, when executed by a processor of a vehicle wheel alignment system, causing the processor to determine alignment parameters for the vehicle, the alignment system comprising: a target capable of being fixedly attached to a wheel of the vehicle; and a camera for observing the target and capturing image data of the target, the determination process comprising: Receive the image data from the camera; The coordinates of the virtual axle point are determined at least in part based on image data of the target captured as the vehicle moves, causing the wheels and the target to rotate. The wheel rotation axis is calculated at least in part based on image data of the target captured as the vehicle moves, causing the wheels and the target to rotate; and The positioning parameters for the vehicle are calculated at least in part based on the coordinates of the virtual axle point and the wheel rotation axis. 19. The non-transitory computer-readable medium of claim 18, comprising: calculating axle points based at least in part on captured image data, and using the axle points to calculate positioning parameters for the vehicle. 20. The non-transitory computer-readable medium of claim 18, wherein the determination process further comprises: The rolling yaw of the wheel is calculated at least in part based on the captured image data; and Rolling yaw calculation is used to calculate the positioning parameters for the vehicle. 21. The non-transitory computer-readable medium of claim 18, wherein the determining process further comprises: using a nonlinear least squares iterative method to calculate the coordinates of the virtual wheel axle point. 22. The non-transitory computer-readable medium of claim 21, wherein the nonlinear least squares iterative method comprises one of the Nelder-Mead simplex algorithm, the Levenberg-Marquardt algorithm, and the gradient descent algorithm. 23. The non-transitory computer-readable medium of claim 18, wherein the determining process further comprises: using a grid search algorithm to calculate the coordinates of the virtual wheel axle point. 24. The non-transitory computer-readable medium of claim 18, wherein the target is attached to the wheel such that the origin of the target is offset from the axis of rotation of the wheel, or such that the origin of the target is substantially arranged on the axis of rotation of the wheel. 25. The non-transitory computer-readable medium of claim 18, wherein the determining process further comprises: comparing the calculated virtual wheel axle point coordinates with a predetermined range of reference virtual wheel axle point coordinates for the target, and notifying the user when the calculated virtual wheel axle point coordinates exceed the range of reference virtual wheel axle point coordinates.