WHEEL ALIGNER WITH ADVANCED DIAGNOSTICS AND NO-STOP POSITIONING
1. A vehicle wheel alignment system comprising:
- a plurality of cameras, each camera for viewing a respective set of a plurality of three-dimensional points on a respective wheel of the vehicle and capturing image data of the set of points as the wheel is continuously rotated a number of degrees of rotation without a pause,wherein the image data is used to calculate a minimum number of poses of the wheel;
wherein each of the minimum number of poses of the wheel is stored;
wherein the stored minimum number of poses of the wheel comprises at least one pose for every five degrees of rotation captured by each camera as the wheel is continuously rotated the number of degrees of rotation without a pause; and
wherein at least one of the cameras comprises a data processor for performing the steps of;
preprocessing the image data; and
calculating an alignment parameter for the vehicle based on the preprocessed image data.
A vehicle wheel alignment system has a plurality of cameras, each camera for viewing a respective target disposed at a respective wheel of the vehicle and capturing image data of the target as the wheel and target are continuously rotated a number of degrees of rotation without a pause. The image data is used to calculate a minimum number of poses of the target of at least one pose for every five degrees of rotation as the wheel and target are continuously rotated the number of degrees of rotation without a pause. At least one of the cameras comprises a data processor for performing the steps of preprocessing the image data, and calculating an alignment parameter for the vehicle based on the preprocessed image data.
- 1. A vehicle wheel alignment system comprising:
a plurality of cameras, each camera for viewing a respective set of a plurality of three-dimensional points on a respective wheel of the vehicle and capturing image data of the set of points as the wheel is continuously rotated a number of degrees of rotation without a pause, wherein the image data is used to calculate a minimum number of poses of the wheel; wherein each of the minimum number of poses of the wheel is stored; wherein the stored minimum number of poses of the wheel comprises at least one pose for every five degrees of rotation captured by each camera as the wheel is continuously rotated the number of degrees of rotation without a pause; and wherein at least one of the cameras comprises a data processor for performing the steps of;
preprocessing the image data; and
calculating an alignment parameter for the vehicle based on the preprocessed image data.
- View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 21, 22)
- 11. A vehicle wheel alignment system comprising:
wherein the image data is used to calculate a plural minimum number of poses of the wheel as the wheel is continuously rotated the number of degrees of rotation without a pause; and wherein at least one of the cameras comprises a data processor for performing the steps of preprocessing the image data, and calculating an alignment parameter for the vehicle based on the preprocessed image data.
- View Dependent Claims (12, 13, 14, 15, 16, 17, 18, 19, 20, 23, 24)
The present application claims priority to U.S. Provisional Application No. 62/238,017, entitled “Portable Vehicle Aligner With High Speed Camera Based Processing and Advanced Analytics,” filed Oct. 6, 2015, which is incorporated herein by reference in its entirety.
This invention relates to an automotive wheel aligner of the machine vision type.
Conventional camera-based 3D machine vision-type aligners are not very portable. Camera based wheel aligners have a pair of cameras mounted such that each camera can look down a side of the vehicle at targets attached to the wheels. The cameras are usually mounted about 100″ apart and are about 100″ in front of the vehicle. Also, the vehicle needs to be raised so the alignment technician can work underneath it. Thus, the aligner cameras either have to move up and down with the rack, or be mounted high enough to be able to see the targets when the rack is raised. The cameras can be over 100″ off the ground when raised. Conventional aligners also include a console that may house a monitor, printer, computer, and target and clamp storage. Many camera based aligners'"'"' cameras are bolted to the ground and cannot be moved from one alignment bay to the next. However, it is useful to be able to move the aligner to another bay, because the vehicle under test might require parts, and removing it from the rack is not feasible. In such a case, the aligner is no longer useable. It is therefore highly desirable to be able to move it to a new location.
There are some camera based aligners that are movable; however, these have some significant limitations. In one of these aligners, the camera beam is simply bolted to the side or back of the console. This makes for a very large assembly that needs to be rolled together through the shop. This assembly is over a 100″ wide, 36″ deep and at least 80″ high. There is no easy way to maneuver this aligner in what typically is a crowded shop environment. Another portable camera based aligner has separate poles for the cameras and the console, so the technician has to move three discrete components to the new location. Also, the cameras are mounted on poles that are over 100″ tall, and they are difficult to move through doors or other areas with a low ceiling. There are also other types of camera aligners that mount to the rack. These are smaller since there is no pole; however, there are special mounting requirements for each rack. It is clear that a better solution to the portable camera based aligner is desirable.
Machine vision vehicle alignment systems using movable cameras and targets attached to vehicle wheels, also known as “image aligners,” are well known. The targets are viewed by the cameras such that image data obtained for a prescribed alignment process can be used to calculate vehicle alignment angles for display through a user interface, usually a computer monitor. Early system implementations included rigid beams that connected the cameras so that their position and orientation with respect to each other could be determined and be relied upon as unchanging. Later system implementations were introduced comprising the use of cameras not rigidly connected to each other, but using a separate camera/target system to continuously calibrate the position of one vehicle mounted target viewing camera to another. This type of system is described in U.S. Pat. Nos. 5,535,522; 6,931,340; 6,959,253; and 6,968,282, all of which are hereby incorporated by reference herein in their entirety. An example of a vehicle wheel aligner using such image processing is the Visualiner 3D or “V3D”, commercially available from John Bean Company, Conway, Ark., a division of Snap-on Incorporated
There are many factors that influence the measurement of a vehicle that can lead to a good aligner and a good mechanic to get bad readings. An uneven roll surface during the roll back procedure can cause the wheel to move or rotate about a different axis than the one it is rolling about. Camera based aligners require measurements of alignment parameters, which include an axis of rotation, a wheel angle, a vehicle wheel plane, or a vehicle wheel center. Another influence in bad readings is if, during caster swing, the vehicle brakes are not locked and the vehicle rolls, or the skid plates and turntables are not unlocked or are sticky, then large forces can build up within the vehicle and cause a bad reading. Also, if during adjustment the skid plates or turn tables are sticky, there is a buildup of forces within the vehicle that adversely affect the alignment. Other areas of contribution to reduced accuracy is if the rack is twisted, or the vehicle is jacked up and not let down and settled correctly.
There is a need for an aligner to incorporate cameras that can take measurements fast enough and continuously to measure, detect, and correct or warn of a problem with the alignment measurement. Further, there is a need for an aligner that measures, detects or corrects all possible issues that might make the alignment readings incorrect. With such an aligner the technician can be confident that the alignment just performed is correct.
Current camera based aligners require a positioning or runout procedure to measure the wheel axis (also referred to as “wheel spindle” or “axis of rotation” throughout this document), and a steering swing procedure to measure the caster and steering axis inclination (SAI). This procedure is time consuming and frequently requires the technician to hold the vehicle steady at certain points of the positioning and caster swing.
There is a need for a faster positioning procedure and a faster caster swing procedure, where the mechanic can perform the procedure without stops or delays.
When an aligner is in need of repair, a service technician is sent to the site. Sometimes the issue could be solved if the service technician could operate the aligner remotely, avoiding a costly service call. Methods exist to take control of an aligner using special software and an internet connection, but there are limits to their capabilities.
There is a need for a remote display/interface for the technician on site using the equipment, and a remote display/interface that service personnel at a different location can use to control the aligner, all on a simple readily available device. A solution that easily does both of these in one simple architecture would be highly desirable.
Given the foregoing, the ideal camera based aligner is self-contained on a single console with a height and width to navigate through the typical shop. Shops are laid out so mechanics have the ability to maneuver easily from area to area, and thus it is desirable that the self-contained aligner fits through these areas. Accordingly, the disclosed aligner incorporates the following features in several embodiments:
- Self-calibrating cameras attached to arms that fold, to reduce the width of the aligner.
- Folding arms movably attached to the console to minimize the width of the console, resulting in a single self-contained aligner and console.
- A locking mechanism to hold the arms closed, yet have easy unlocking for regular width alignment.
- A screw shaft and mechanism to lower and lock the arms, and to lower the center of gravity and overall height of the aligner to facilitate less tipping and pass-way through low ceilings or doorways. Vertical movement with the screw shaft is reduced or eliminated.
- Various methods of automatically closing the arms when they are lowered for easier storage and transportation. For example, two posts on the side of the console that push up the arms, an internal extra slide within the slide car that pulls the arms closed, another mechanical motor and mechanism to pull the arms closed.
- Wheel clamps/targets mounted on the console at an angle to minimize the depth and width of the console.
- Clamshell “structural” arms to reduce the weight and cost of the arms, thereby allowing for a small torque motor and lower power requirement, thus further reducing weight and costs.
Self-contained portable aligners can advantageously be moved from one location or alignment bay to the next with ease. However, they typically lack functionality once they are located at the alignment bay. When a car is being aligned it is desirable to be able to see the readings from one side or the other side of the vehicle. Self-contained portable aligners have all the equipment for measurement in a single easy to move design, but if desired it is difficult to move just the display device to the side of the vehicle so the technician can see it. Being all in one, if the technician moves the console for easy viewing of the display, he has also moved the cameras and they can no longer see the wheel targets. There is a need for a portable aligner that once in the bay could also be portable to be able to view the display device. The disclosed aligner addresses the foregoing problems by incorporating the following features in several embodiments:
- A nesting console with a portable camera base.
- Wireless communication between the cameras and the console.
The ideal camera based aligner also has a rapid camera measurement device to measure the discrete elements that make up certain critical alignment procedures, in order to measure, detect, correct, or warn of problems. It also has the ability to perform procedures to measure the vehicle wheel axis and the caster and steering axis inclination (SAI) continuously, without pausing or holding. Accordingly, the disclosed aligner incorporates the following features in several embodiments:
- A rapid camera measurement device to measure the discrete elements that make up the positioning procedure to measure, detect, correct, or warn of issues.
- A rapid camera measurement device to measure the discrete elements that make up the caster swing procedure to measure, detect, correct, or warn of issues.
- A rapid camera measurement device to measure the discrete elements that make up the adjustment of the vehicle to measure, detect, correct, or warn of issues.
- A rapid camera measurement device to measure the discrete elements that make up the jacking and settling of the vehicle to measure, detect, correct, or warn of issues.
- Thrust Angle checking before and after procedures that would indicate the alignment rack'"'"'s turntables or skid plates are sticking.
- Rack twist measurement correct or warn.
- Detection of suspension changes during alignment or measurement procedures to correct and/or warn.
- Continual rapid rigid body analysis of measurements to detect stress build ups, and warn or correct.
- Non-stop runout procedures and caster swing procedures.
The disclosed aligners also include a fully networked system aligner (web server) for complete, simple, inexpensive functionality of remote display and control of the aligner from the user'"'"'s shop or from the service desk many miles away.
In an embodiment of the present disclosure, a vehicle wheel alignment system comprises a plurality of cameras, each camera for viewing a respective target disposed at a respective wheel of the vehicle and capturing image data of the target as the wheel and target are continuously rotated a number of degrees of rotation without a pause. The image data is used to calculate a minimum number of poses of the target of at least one pose for every five degrees of rotation captured by each camera as the wheel and target are continuously rotated the number of degrees of rotation without a pause. At least one of the cameras comprises a data processor for performing the steps of preprocessing the image data, and calculating an alignment parameter for the vehicle based on the preprocessed image data.
Further features and advantages of the present disclosure, as well as the structure and operation of various embodiments of the present disclosure, are described in detail below with reference to the accompanying drawings. It is noted that the present invention is not limited to the specific embodiments described herein. Such embodiments are presented herein for illustrative purposes only. Additional embodiments will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein.
The accompanying drawings, which are incorporated herein and form part of the specification, illustrate the present invention and, together with the description, further serve to explain the principles of the present invention and to enable a person skilled in the relevant art(s) to make and use the present invention.
Additionally, the left-most digit of a reference number identifies the drawing in which the reference number first appears (e.g., a reference number ‘310’ indicates that the element so numbered is first labeled or first appears in
Further embodiments, features, and advantages of the present invention, as well as the operation of the various embodiments of the present invention, are described below with reference to the accompanying figures.
While embodiments described herein are illustrative embodiments for particular applications, it should be understood that the invention is not limited thereto. Those skilled in the art with access to the teachings provided herein will recognize additional modifications, applications, and embodiments within the scope thereof and additional fields in which the invention would be of significant utility.
The embodiments described herein are referred in the specification as “one embodiment,” “an embodiment,” “an example embodiment,” etc. These references indicate that the embodiment(s) described can include a particular feature, structure, or characteristic, but every embodiment does not necessarily include every described feature, structure, or characteristic. Further, when a particular feature, structure, or characteristic is described in connection with an embodiment, it is understood that it is within the knowledge of one skilled in the art to affect such feature, structure, or characteristic in connection with other embodiments whether or not explicitly described.
Self-Calibrating Aligner with Improved Portability
An aligner with certain discrete components that can be fully disassembled and assembled in any configuration is shown in
All of the components can be assembled on common base 16m and be an all-in-one portable aligner (
The disclosed 3D visual aligner includes a three-camera self-calibrating measuring system having a first camera pod 16g and a second camera pod 16h. First camera pod 16g comprises a forward-facing first camera 16p for viewing a pair of targets (such as the targets T of clamps 16d) when the targets T are attached to the left-side wheels of the vehicle to be aligned, and a calibration camera 16s extending from the rear of the pod 16g and facing the second camera pod 16h. Second camera pod 16h comprises a forward-facing second camera 16r for viewing a pair of targets (such as the targets T of clamps 16d) when the targets are attached to the right-side wheels of the vehicle to be aligned, and a calibration target 16q. The two forward facing cameras 16p, 16r are not rigidly connected to each other; that is, their relative position and orientation can change. The calibration camera 16s is rigidly attached to the forward facing “Left” camera 16g, and views the calibration target 16q that is rigidly attached to the forward facing “Right” camera 16r. The calibration camera 16s is used to measure the relative orientation of the forward facing cameras 16p, 16r in a conventional manner, as explained, e.g., in U.S. Pat. No. 6,968,282, incorporated by reference herein.
Camera pods 16g and 16h are supported by reversible arms 16n. Reversible arms 16n are attached to moveable carriage 17b by removable pins 17c (
In certain embodiments, all the major components of the disclosed aligner, including the two camera pods 16g, 16h, the display device, the printer 16j, and any other I/O device such as a keyboard and mouse, communicate with each other wirelessly. In such embodiments, a conventional wireless communication “hub” which controls the wireless communication is located at one of the components; e.g., one or both of the camera pods 16g, 16h. Also, the aligner can be connected to an external network (e.g., via the wireless communication hub) and communicate therewith, as explained in detail herein below.
An exemplary embodiment is shown in the system block diagram of
The cameras and camera circuit boards in camera pods 2201, 2202 facing the wheel targets are identical, and both perform image processing. In certain embodiments, alignment calculations are also performed by the camera boards of pods 2201, 2202, and the results are communicated to the mini PC 2204, which serves a user interface and displays the alignment results. Note that the terms “camera” and “camera board” are used interchangeably throughout this document, as those of skill in the art will appreciate. The calibration camera 2201a mounted in the LH camera pod 2201 that looks across at the calibration target 2202a mounted in the RH camera pod 2202 does no image processing and, as a result has a simplified board. The image processing for the calibration camera 2201a is done by the board for the LH camera 2201b. There is also a drive-on camera 2202c mounted in the RH camera pod 2202 that views the vehicle as it drives onto the lift. It relies on the RH camera 2202b for image processing. In other embodiments, all image processing is performed by one camera board. Alternatively, the camera boards do not perform any image processing, and instead send raw images to the mini PC 2204 for processing.
In other features of embodiments of the disclosed portable aligner, in a “standard” configuration the monitor 16b of the display device, or the display device itself, is attached to column 16f by a bracket 17a (see
Base plate 16e, column 16f, moveable carriage 17b, arms 16n and camera pods 16g, 16h comprise a self-contained measuring system that can be attached to mobile base 16m, to the floor, or to a suitable wall or upright structure. Base plate 16e contains a plurality of holes that can be used to attach base plate 16e to mobile base 16m. This same plurality of holes could be used with floor anchors to attach the measuring system to the floor. Column 16f also has multiple brackets 17a, 17d (see
Referring now to
A wireless connection between the printer 16j and the camera measuring system allows the printer to be installed in certain embodiments in the keyboard shelf 16a, or at any other user preferred location within the alignment shop, or within a nearby office or room.
A wireless connection between the display device, the wireless communications hub, the user input devices such as keyboard and mouse, and/or the printer 16j allows for the addition of a second fully detached console in a further embodiment. This configuration allows a mobile camera measuring system with a separately mobile user input station. In one embodiment shown in
The docking functionality can be achieved by a number of well-known techniques. In one embodiment, mechanical latching/connecting device(s) secure the satellite console 2300 when docked. This mechanical device allows for simple and quick docking and undocking of the satellite console 2300 to the main console 2310. Another way of implementing the docking functionality is a magnetic mechanism 2330 to allow for simple and quick docking and undocking. This mechanism 2330 can be an electromagnet that is engaged/disengaged with the press of a button. Alternatively, it can be a permanent magnet that is engaged by simply placing the satellite console base 2300a close enough to the main console base 2310a to engage the magnet 2330. To disengage the permanent magnet 2330, a foot lever or other similar mechanism can be used to break the magnetic attraction between the two consoles 2300, 2310.
When under a vehicle making alignment adjustments, it is difficult if not impossible to see the main display. A solution for alleviating the viewing disadvantages of having the main display at the front of the vehicle is to make the display height adjustable on the console, as shown in
Aligner with Advanced Analytics
Embodiments of an aligner having advanced analytics will now be described, which depends on high-frequency data collection and continuous or semi-continuous analysis of measurements. The disclosed advanced analytics determine whether certain measurements are indicative of operator error and/or equipment faults. The aligner uses the advanced analytics to inform the operator of the error or fault and/or correct for the error or fault. This functionality is enabled by the aligner'"'"'s very fast measurement process which; for example, calculates multiple target poses for the vehicle wheels as they are moved, using optimized iterative algorithms to estimate the position and orientation of the reference targets. See, e.g., U.S. Pat. Nos. 5,535,522; 7,855,783; 6,871,408; and 7,069,660; and U.S. Patent Application Publication 2016/0195388, hereby incorporated by reference herein.
Arrow 30 of
The disclosed aligner further comprises a data processor (not shown), such as a conventional personal computer (PC), having software with instructions to cause the data processor to perform the calculations described herein electronically.
The method and apparatus described herein is also applicable for use with a hybrid aligner system similar to that described in U.S. Pat. No. 7,313,869, which is hereby incorporated by reference in its entirety, and its continuation patents.
A key technology enabling the disclosed aligner and the functionality described herein is a very fast measurement process. For this innovation, the core measurements of interest are the poses (positions and orientations) of targets that are rigidly mounted to the vehicle wheels. Performing a fast measurement process thus equates to performing measurements of target pose very rapidly. In imaging aligners, computing pose rapidly involves performing optimized image processing and applying optimized iterative algorithms to estimate the position and orientation of the reference targets. The high speed measurement process provides for many updates and checks to be performed during the course of a wheel turning processes which may only take several seconds. To measure the pose of wheel mounted targets from individual cameras, such cameras 10L and 10R of
Examples of well-known camera calibration processes are the Heikkila method; Zhang'"'"'s method; the Faugeras-Luong method; the Hartley-Zisserman method; and the Triggs method. To compute the 3D pose of a target from a single 2D camera it is further required to have knowledge of the geometry of the target that is being observed. With knowledge of the target geometry and knowledge of the internal camera geometry, it is possible to compute the 3D pose of that target based on a single 2D image. An example of a commercially available “fast” camera usable to implement the disclosed aligners is the VC Z series camera available from Vision Components GMBH of Ettlingen, Germany.
The process of computing target pose is conventional, and starts with acquiring an image of the wheel mounted target. This image is then processed to identify image feature points that correspond to reference fiducials in the target. These reference image points are then associated with reference fiducial points in the target. Finally, an iterative pose estimation process is performed. Pose estimation answers the question, “Where does a target need to be located, and how must it be oriented, to produce the pattern of image feature points that I measured?”
In the disclosed aligners, the entire measurement process (image acquisition, image processing, image feature point to target fiducial point correspondence, and pose estimation) is performed repeatedly in a loop for all targets used until target pose information is no longer needed for the alignment process. To acquire measurements very rapidly (for example, greater than 20 poses per second) it is necessary to perform all four steps rapidly. The data processing steps must be implemented with highly efficient algorithms and they must also be implemented on processors with architectures that are optimized for the image processing and numerical linear algebra steps used in the data processing algorithms. Examples of well-known processors with architectures optimized for image processing and numerical linear algebra include DSPs (Digital Signal Processors); GPUs (Graphics Processing Units); FPGAs (Field Programmable Gate Arrays); and ASICs (Application Specific Integrated Circuits). Examples of highly efficient data processing algorithms include Gaussian filtering; gradient descent optimization; Sobel edge detection; Canny edge detection; SURF feature point detection; and optical flow point tracking. An example of a commercially available processor with architecture optimized for the application of this disclosure is Model TDA2x ADAS DSP/ARM System-on-Chip processor, available from Texas Instruments of Dallas, Tex.
In certain embodiments, the data collection procedures and advanced analytics described herein below with reference to
The data collection procedures and advanced analytics described herein below with reference to
To calculate the alignment of the vehicle, it is well-known that the aligner must locate the spindle of each wheel. To locate the spindle, the aligner typically measures the pose of the target attached to the wheel in at least two places before and after the wheel is rotated. Wheel spindle calculations are well-known, and exemplary algorithms described for such calculations are described in U.S. Pat. Nos. 5,535,522 and 6,148,528. To locate the spindle accurately, the vehicle must roll around a single axis. Any variation from a single axis roll will cause errors. This process is commonly referred to as runout compensation or rolling runout compensation. Using fast processing cameras as discussed herein above, the disclosed aligner can take discrete measurements at many places during the rolling procedure. With these measurements, the disclosed aligner can take many axis calculations, as well as other calculations looking for potential errors and solutions. Some of these errors might be a bump in the path of the vehicle'"'"'s roll, a slide of the skid plates, or multiple axes of rotation; e.g., an inadvertent turning of the steering wheel, or a wheel wobble due to bad bearings. See, e.g., U.S. Patent Application Publication 2016/0195388.
There are other useful well-known calculations that can be made during the rolling process; for example, the center of the wheel can be calculated by looking at the path and pose of the targets through the rolling process. This can be then used to establish the vehicle reference plane, or vehicle dimensions.
These advantageous fast measurements are not restricted to the cameras looking at the wheels; the calibration camera (such as camera 20 of
In typical imaging wheel alignment systems, one acquires images of vehicle wheels with clamps that are rigidly mounted to each individual wheel rim, each clamp carrying a target that is viewed by one of the cameras. Examples of such clamp/target assemblies are target/clamp 80a, 88 of
The present disclosure addresses this problem by rapidly acquiring and recording measurements of wheel-mounted target pose (i.e., target position and orientation) throughout the wheel rolling process and analyzing these measurements to check the wheel alignment assumptions.
The process in which data is collected by the disclosed aligner is depicted in
Once the compensation data acquisition process completes, the poses logged for all four wheels are then passed into through a series of diagnostics checks in a “Compensation Analytics Engine” shown in
One set of inputs for the Compensation Analytics Engine of
TST vectors can be computed using a number of well-known techniques. One technique employs a “3D coordinate measurement machine,” wherein the wheel target comprises suitably positioned reference fiducial markers. These points are measured and then used to define the origin point and coordinate axes of the target coordinate system. An additional set of 3D coordinates are obtained from tips of the three or four symmetrical wheel clamp “claws” of the clamp that holds the wheel target on the wheel. In a conventional self centering clamp, such as clamp assembly 16d shown in
As is clear from
One of the assumptions used when performing compensation is that the surface over which the vehicle rolls is flat. In cluttered real world scenarios, this may not necessarily be the case. As those of skill in the art will appreciate, the alignment lift can possess large discontinuities in its surface. Using the large quantity of data collected during the compensation roll in the disclosed aligner, these surface continuities can be detected and in some cases corrected for. An embodiment for performing the surface flatness checks at steps 204-212 of
The basic principle in the surface flatness detection algorithm is to employ the fact that the center of a wheel rolling on a flat surface moves in a straight line. In this embodiment, the center of the wheel is computed by applying the TST (Target Spindle Translation) vector to each target pose that was saved as part of the data acquisition process in
A more complicated situation arises when the best fit line through the lift surface does not have a sufficiently low RMS fit error at step 306. In this case, the iterative outlier removal process is started (the bottom box inside of
This process of steps 310-316 iterates until either the best fit line achieves a sufficiently low RMS error or until an excessive number of data points are removed, which is checked at step 320. The bare minimum number of 3D points required to uniquely specify a 3D line is two. Using only two points is inadmissible for this test, however, as two points will always form a perfect best fit line with no error whatsoever (defeating the purpose of this test). In practice, the minimum number of measurements required is at least 6.
If outlier measurements are removed and the RMS best fit line error is sufficiently low, then only those poses corresponding to the inlier wheel center point measurements are used to compute the wheel axis. In this case, the inconsistent rolling surface is corrected for and the process progresses to the next check.
If too many outlier measurements are removed, then the non-uniform rolling surface cannot be compensated for. The surface flatness check fails and the user is alerted at step 322 that the observed surface non-uniformity couldn'"'"'t be corrected. In such a scenario, the runout compensation process cannot be completed successfully (step 324).
There are various ways in which surface flatness checks could be implemented. In the above-described embodiment, reference was made to the RMS error of a best fit line fitted through the 3D wheel center points. One could use additional optimization metrics other than the RMS error. For example, one could use changes in the slope of the best fit line as the criteria to minimize other than the RMS best fit error.
Wheel alignments are performed by the disclosed aligner on a conventional alignment lift, a specialized piece of equipment with movable plates under the vehicle wheels (“turn plates” for the front wheels and “slip plates” for the rear wheels). The plates can be set by the technician in either a locked or an unlocked state. When locked, the plates do not translate when forces are applied to the surface. When unlocked, the plates translate freely when a force is applied. This allows for vehicle wheels to translate and rotate without inducing additional stress in the vehicle suspension. During certain wheel alignment procedures such as runout compensation, however, it is necessary for the plates to be locked and to not translate freely when the vehicle rolls over them. Free translation of a wheel means that it is free to translate without a proportional rotation about the wheel axis. Free translation of wheels without corresponding rotation can cause errors in runout compensation.
The process for detecting plate slipping at steps 214-222 of
As the vehicle rolls, for each wheel there must be an elapsed angle of rotation that is proportional to the linear travel distance. The linear travel distance is the distance translated by the center of each wheel. Formula 1 below shows the ideal proportional relationship between wheel center translation and elapsed angle of wheel rotation (in radians). This proportional relationship is used to perform a linear least squares best fit of wheel translation distance vs. angle of rotation. Errors of predicted vs. measured rotation are then used to detect and in some cases correct for observed instances of translation without rotation.
dtrans∝θrot Formula 1: Ideal Wheel Travel Distance and Elapsed Angle of Rotation
As is schematically depicted in
As shown in
Referring now to
If the RMS error is not below this threshold then an instance of translation without rotation has been detected. The iterative outlier removal process then runs. At each iteration, the worst measurement (with the largest absolute error in the best fit line) is removed (step 410). Translations and rotations are recomputed (taking care to adjust input poses for the omitted measurement) and the translation vs. rotation best fit line is recomputed (step 412) along with fit errors for the remaining data points (step 414). If the RMS error is sufficiently low after omitting this worst point (step 416) then the process completes (steps 418, 408) and runout compensation proceeds to the next step. This process of omitting the worst remaining point of steps 410-416 iterates until either a sufficiently good best fit line RMS error is achieved or until too many points have been removed (this is checked at step 420), at which time the user is alerted at step 422 that the observed plate slipping problem couldn'"'"'t be corrected, and the plate slipping check fails (step 424).
In the above-described embodiment of plate slipping detection/correction, the criteria for detecting/correcting plate slipping is a proportionality check between wheel center point translation vs. wheel rotation. With time-tagged pose data, in further embodiments the elapsed angle of rotation is interpolated to all measurement times on all wheels. With angle of rotation and time, angular velocity can be computed. If any time during the rollback, one wheel shows a much lower angle of rotation than the other wheels under observation, then this information could be used to detect plate slipping.
The principle quantities of interest measured during the compensation roll are the wheel axes. The wheel axes are the (virtual or real) spindles about which each wheel rolls. It is assumed during the compensation process that all measured rotation occurs about wheel axes. This might not be true, however. If a wheel does not closely approximate a circle during a roll then there will not be a unique wheel axis of rotation. For example, if a tire is somewhat flat or if a rim is bent, the wheel may exhibit wobble as it rolls.
When a wheel possesses a significant amount of wobble, the true wheel axis of rotation cannot be measured directly from changes in the orientation of the wheel mounted targets. Rather, the measured axes of rotation will precess about an axis. This “axis of precession” is the true wheel axis of rotation. This phenomena is illustrated in
The process for computing the axis of precession (i.e., the true wheel axis) from individual measured axes of rotation is depicted in the flow chart of
Accordingly, as shown in
When the standard deviation of the wobble angles is below the error threshold, the wheel axis precession is compensated for by assigning the axis of precession to be the true wheel axis for a following alignment step (step 850). When the standard deviation of the wobble angles is above the error threshold, the user is alerted that the system cannot correct for the wheel axis precession (step 860).
The algorithm in which axis of precession is computed is described in
Accordingly, computing the axis of precession comprises inputting the set of all measured wheel axes of rotation found in step 810 (step 900), and computing a best fit plane through the tips of the calculated wheel axis vectors (step 910). The root mean square (RMS) planarity error is then computed at step 920, and it is determined whether a deviation from the best fit plane is below a predetermined RMS error threshold at step 930.
When the deviation from the best fit plane is above the RMS error threshold, the wheel axis precession is computed by removing the wheel axis of rotation vector with the largest deviation from the best fit plane (step 950), recomputing the best fit plane (step 910), recomputing the RMS planarity error (step 920), determining whether the deviation is below the RMS error threshold (step 930), and repeating steps 950, 910, and 920 until the RMS deviation from the best fit plane is below the error threshold or until less than a predetermined minimum number of data points remain (this is checked at step 960). If less than the minimum number of axes remain, the user is alerted that the system cannot find a valid axis of precession, and therefore cannot correct for wheel axis precession (step 970). On the other hand, if the deviation is found to be below the RMS error threshold at step 960, and more than the minimum number of data points remain, the normal of the best fit plane is assigned to be the axis of precession at step 940.
One alternative way in which wheel axis of rotation precession could be detected is by stopping the vehicle at multiple points during the compensation roll and collecting data at each of these stopped positions. Axes of rotation could be computed for poses collected at each stopped position and the axis of precession could then be computed from these axes of rotation in an analogous manner to the previously-disclosed embodiment.
When rolling the vehicle during runout compensation, it is assumed that the measurement cameras are static. If the measurement cameras are bumped or otherwise moved during the data acquisition of
In this disclosed embodiment, there are two forward facing cameras that are not rigidly connected; that is, their relative position and orientation can change. For example, such a camera arrangement is shown in
An embodiment for performing the camera instability checks at steps 234-242 of
Accordingly, detecting instability of one or more of the forward facing cameras comprises starting the compensation procedure (step 1000) by measuring an initial calibration target pose in that camera'"'"'s coordinate system (step 1005). As the vehicle rolls (step 1010), a new calibration target pose is acquired (step 1015). At steps 1020-1030, if the new target pose does not correspond to the calibration target (i.e., it corresponds to a wheel mounted target), the pose is used as input to the larger runout compensation process, and the amount of wheel rotation is checked to determine whether all four vehicle wheels have rotated an adequate amount to complete the compensation procedure.
At step 1035, the initial calibration target pose is compared with the succeeding one of the calibration target poses, to determine whether it deviates from the initial calibration target pose more than a threshold amount (step 1040). The compensation procedure is stopped (step 1045) and the user is alerted that the cameras are unstable when one of the succeeding calibration target poses deviates from the initial calibration target pose more than the threshold amount (step 1050). Otherwise, the process is repeated until the wheels have rotated an adequate amount.
Differences in the calibration target pose are computed for both rotation and translation. As a representative threshold, new calibration target poses must be within a translation of 0.15″ and a rotation of 0.15° from the initial stored calibration target pose.
Through this process, data is acquired for performing runout compensation and for performing the various analytics checks described elsewhere. The camera instability process is run online during the compensation data acquisition process. If at any time the calibration target pose is measured to be too different from the reference target pose, the user is alerted that the camera instability is too large and the procedure is stopped.
There are various other ways in which detection of camera instability during compensation could be implemented. One could use additional sensors. For example, if there were accelerometers rigidly attached to each forward facing camera, one could monitor the accelerations to determine when camera pods change their pose. Additionally, camera instability detection can be implemented by securing a fixed reference target (or a series of such targets) to a rigid structure while still remaining in the field of view of forward facing cameras. For example, the targets could be secured to a wall or to a post on the floor. The pose of the reference targets could be measured throughout the compensation process. If the fixed reference target pose changes, as measured by any of the forward facing cameras, then it can be readily determined if the static camera assumption made during compensation is valid or not.
The aligner measures the pose of the targets attached to the wheels and calculates the alignment of the vehicle after every pose measurement. When the user is making adjustments to the vehicle or performing procedures on the vehicle, he could make mistakes, do procedures incorrectly, or skip procedures all together, resulting in an inaccurate alignment. By taking fast measurements, the disclosed aligner can look for and detect when the operator makes one of these mistakes and inform him earlier, to prevent unwanted delay and give him quick feedback to help him learn and prevent future error repetitions.
A recognized source of measurement error when performing vehicle wheel alignments is stress changes induced in the vehicle suspension. If a vehicle suspension is stressed beyond typical rest conditions, the stress tends to gradually release as the vehicle is moved/adjusted during the course of an alignment procedure. This release of the suspension stress can significantly change wheel alignment angles. This gradual changing of wheel alignment angles can be particularly harmful in procedures like runout compensation. If a vehicle suspension de-stresses during runout compensation, the runout compensation could be significantly incorrect and as a result vehicle wheel alignment angles could be adjusted to incorrect values. This could in turn cause premature tire degradation and/or result in sub-optimal driving comfort.
There are several ways vehicle suspensions could become prematurely stressed in the course of performing alignment angle measurements. It is common for alignment technicians to raise vehicles up on alignment lifts to inspect the vehicle undercarriage prior to performing any measurements or adjustments. If the vehicle is lowered to the alignment lift surface and the slip plates are locked, the vehicle suspension can become stressed as the weight of the vehicle becomes supported by the suspension and the suspension is not free to adjust to its resting position. Another common way in which vehicle suspensions can become stressed is if the vehicle experiences a sharp turn immediately prior to driving onto the alignment lift. The sharp turn induces stress into the suspension and the very short straight line transit distance (essentially the length of the alignment lift) does not provide enough opportunity for the suspension to release.
Whatever the cause, the gradual release of suspension stress during the course of an alignment is a problem that if not detected will negatively impact the alignment process. There are various ways in which the suspension stress change can be detected. One embodiment is depicted in
The data collection loop is depicted in
Once the data collection has finished for all four wheels (i.e., once each wheel has experienced enough elapsed rotation), a suspension stress detection algorithm can be run on the acquired target pose measurements. The computations and decision logic for checking if this suspension change occurred is depicted in
As shown in
Pose measurements for each wheel might not be acquired in a time synchronized manner. For such a scenario, when computing the VCS from the wheel rim center points it is essential to interpolate the wheel center points to all pose measurement times. These interpolated wheel center points are then used to compute the VCS at all measurement times. See steps 1115 and 1120.
A key consideration of this disclosure is that the suspension stress check is performed “behind the scenes,” such that the end user isn'"'"'t burdened with additional tasks or with unnecessary information. The user is only given instructions/notifications in the event suspension stress is detected during the normal course of events performed as part of an alignment process. This decision logic is depicted in the lower right-hand rectangle inside of
A key enabling technology for this steady state monitoring is a very fast measurement process. In
The high speed measurement process provides for many updates and multiple checks (approximately 20 poses collected per wheel during a fast vehicle roll) to be performed during the course of measurement processes which may only take several seconds. The very fast measurement process allows the steady state suspension stress monitoring to be run in the background, not providing any updates to the user unless a problem is detected. This provides for a very user-friendly method to identify measurement problems.
Once a change is discovered, it would be helpful if instead of a warning, the aligner compensated for the change and continued on with the alignment with the corrected readings. To do this, in certain embodiments the measurement change of specific alignment readings of the vehicle is plotted on a graph or table or a formula for a given procedure. Using these empirically derived reference plots/table/formula, the aligner matches where the current vehicle under test is in the settling process and predicts where it will be once completely settled. Thus, instead of alerting the user to a suspension stress change in step 1145, the system could compare the change to an empirical reference value or use the change in an empirical formula to predict a value for the alignment parameter when the suspension is settled, and use the predicted value in the subsequent alignment step. As an example, if a vehicle is under stress at the beginning of the roll back procedure, one way to remove the stresses is to roll it backwards and forwards 3 or 4 times. From empirical testing it is noted that after the first roll back only 50% of the stress is relieved. Therefore, when the current vehicle under test has a change of say 0.1 degree after the first roll, it is predicted that if it were rolled backward and forward 3 more times to its fully settled state, it would change 0.2 degree. This full settlement value could be used to display what that actual reading would be if the vehicle were fully settled.
There are various alternative techniques in which suspension stress changes can be identified. The above-described embodiment of
In the embodiment of
Repeated reference is made to a Vehicle Coordinate System in
The disclosed suspension stress change monitoring is performed during the runout compensation process. This is done out of convenience as the user performs this process regardless of whether or not suspension stress is monitored. However, the disclosed suspension stress check does not have to be performed as part of the compensation procedure. It could be performed as a stand-alone procedure (for example, for alignment systems that do not require runout compensation) or it could be performed in any other alignment procedure where the suspension stress metrics should remain invariant.
The vehicle thrust direction is the direction along which vehicle moves. In passenger vehicles, thrust is defined by the relative toe angles of the rear wheels. Thus, when performing measurements on the front wheels the thrust direction shouldn'"'"'t change. By continuously monitoring the vehicle thrust angle during the course of various front wheel measurements, the disclosed aligner can detect problems in the alignment lift without tasking the end user with any additional measurement steps. This functionality is enabled by the aligner'"'"'s very fast measurement process, which calculates multiple target poses for the vehicle wheels as they move.
Equation 2 shows the well-known formula for computing thrust angle in step 1250. Essentially, thrust angle is the average difference between the right rear and left rear toe angles.
The steady state monitoring process of
Where and how might this steady state monitoring of thrust angle be useful? For example, when performing caster swing the front wheels are turned left to right or right to left. During this process, the slip plates on the alignment lift are unlocked and the vehicle is free to translate along the two dimensional surface of the alignment lift. If for example the left rear slip plate is not free to move but the plates for the other three wheels are free to translate, there will be stress induced in the left rear wheel. This stress can manifest itself by twisting the wheel so as to change the toe angle on the left rear wheel without significantly changing right rear toe angle. As per Equation 2 above, such a series of events will result in a change in the thrust angle. In such a scenario the rapid steady state updates of thrust angle of the disclosed aligner (e.g. 20 updates per second) can be used to alert the user that a problem has been detected which could result in a deficient alignment for the vehicle under consideration.
A key enabling technology for this steady state monitoring is a very fast measurement process. In
The high speed measurement process provides for many updates (e.g., approximately 20 poses collected per wheel during a fast vehicle roll) and checks to be performed during the course of measurement processes which may only take several seconds. The very fast measurement process allows the steady state thrust angle monitoring to be run in the background, not providing any updates to the user unless a problem is detected. This provides for a very user-friendly method to identify deficiencies in support equipment.
In the embodiment of
In the embodiment of
In addition, it is not required to perform the aforementioned thrust angle checks at very frequent time intervals. A slower measurement procedure wherein the end user must repeatedly pause the system so that thrust angle (or other invariant metric) can be recomputed and checked against the reference steady state value could be performed to similar effect as the above-described embodiment. It has the disadvantage of requiring more time and more effort of the end user.
The disclosed aligner allows a technician to perform a runout procedure quickly by pushing and/or pulling the vehicle so its wheels roll through an angle of rotation continuously without pausing at turnaround points. This functionality is enabled by the aligner'"'"'s very fast measurement process, which calculates multiple target poses for the vehicle wheels as the vehicle is rolled (e.g., at least one pose for every five degrees of rotation).
A faster, more user-friendly measurement of runout compensation is provided via a procedure hereinafter referred to as “rapid runout.” In rapid runout, the user pushes the vehicle to roll through an angle of rotation. As the vehicle starts rolling, measurements are collected and logged for post-processing until the elapsed angle of rotation is experienced. Each “measurement” consists of the pose (position and orientation) of a target that is rigidly attached to one of the four vehicle wheels. Throughout the roll, measurements are acquired steadily for all vehicle wheels at semi-regular intervals.
If measurements show a significant rotation from the previous stored measurements for their associated wheel then they get stored for post-processing. The non-static compensation data acquisition process is depicted in
The data acquisition process loops until sufficient elapsed rotation (e.g., 35 degrees) has been measured for all vehicle wheels. The salient point is that data is collected as the vehicle rolls without requiring any pauses or decisions from the end user. Data collection stops automatically as well. The process is seamless.
The pose measurements that are collected and monitored consist of translation and orientation components of the observation targets in a reference coordinate system. Pose is computed from image acquisitions of the reference targets on individual wheels.
Data collection is not required to start and stop in a time-synchronized manner for all four wheels. Once a given wheel has been measured to exhibit sufficient rotation, data collection for that wheel stops. Data collection continues for other wheels until they have been measured to experience enough elapsed rotation. Once sufficient rotation has been measured for all wheels, the collected target pose data is passed to a non-static compensation algorithm. The end result of the algorithm is the wheel axis for each vehicle wheel.
This process is depicted in
The wheel axis is used later in the wheel alignment process to compute toe and camber alignment angles. It is thus of paramount importance to measure this quantity as accurately as possible. The basic principle in computing the wheel axis, as shown in
The axis of rotation can be computed at step 1430 using a number of standard well-known techniques. In one embodiment, the 3×3 rotation matrix that rotates the target coordinate axes from one pose measurement to the next is computed. Then, an eigenvector/eigenvalues decomposition is performed on this 3×3 rotation matrix. The eigenvector corresponding to the principal eigenvalue is then assigned to be the axis about which the target rotates, i.e. the axis of rotation.
A key enabling technology for this steady state monitoring is a very fast measurement process, as described herein above. The high speed measurement process provides for many updates and checks to be performed during the course of a vehicle roll which typically only takes several seconds. The very fast measurement process allows for data redundancy which can enable a more accurate measurement of runout compensation. Most importantly, the user does not need to pause and wait while target poses are being acquired.
There are alternative embodiments by which non-static compensation could be achieved. The specific rolling procedure in which data collection occurs is not material to the present disclosure. For example, in some applications it is preferable to roll the vehicle to the back of the alignment lift and then return it to the front without pausing. In others (for example, in so-called “audit” configurations) it may be preferable to perform one roll without a corresponding return to the initial position. In other scenarios it may be preferable to perform several consecutive shorter rolls. In all scenarios the data acquisition process of
In the embodiment of
In the embodiment of
The end result of the disclosed non-static runout compensation process for each wheel is a single averaged wheel axis vector. However, the computation of a mean wheel axis is not strictly required. For example, one could perform additional statistical analysis to select the most typical “median” wheel axis vector. Additional methods could be used to find the resultant wheel axis vector from the set of stored wheel axes. The disclosure does not depend on a statistical combination of individual wheel axes (mean, median, or other). One could simply take the wheel axis to be the axis of rotation from the largest observed angle of rotation in the set of stored poses for that wheel. The key is that wheel axes are computed without requiring the user to pause and hold at any point during the data acquisition process.
The disclosed aligner allows a technician to perform a caster swing quickly by turning the vehicle wheels continuously without pausing at turnaround points. This functionality is enabled by the aligner'"'"'s very fast measurement process, which calculates multiple target poses for the vehicle wheels as they are turned.
A principle quantity of interest in vehicle wheel alignment is caster angle. Caster is defined as the back/forward inclination angle of the steering axes for the front vehicle wheels. To measure this inclination angle of the steering axis it is necessary to make the vehicle front wheels exhibit a rotation about their steering axes. In other words, one must turn the front wheels left and right an appreciable amount to enable an accurate measurement of the steering axes. In typical wheel alignment systems, this process of turning left and right requires a pause at each of the various turnaround points. This pause adds unnecessary delay to the caster angle measurement process.
This unnecessary pause is eliminated in a process hereinafter called “non-static caster”. This measurement process is depicted in
A key enabling technology for this steady state monitoring is a very fast measurement process, as described herein above, which equates to performing measurements of target pose very rapidly. The high speed measurement process provides for many updates and checks to be performed during the course of a wheel turning process which may only take several seconds.
To compensate for the camera and processing not being fast enough, in certain alternative embodiments the aligner collects and stores images and only process every Nth image (for example, every 10th image). This technique results in an effect similar to a non-static caster measurement, in that the aligner can instruct the user to stop turning the wheel as soon as it measures enough rotation about the steering axis without a pause, and then instruct him to turn the wheel back, all the while saving the images of this “burst mode” for future processing. In the background, when bandwidth is available the saved images are processed; however, there would be a delay after the caster swing is complete for all the images to be processed and the caster measured and ultimately displayed to the technician. Burst mode has certain disadvantages because (1) the user does not know as quickly when sufficient rotation has occurred; (2) the LED strobe isn'"'"'t being updated as regularly as when using “fast” processors so the strobe might be too dim or too bright, producing reduced quality measurements; (3) a significant amount of fast data storage is required; and (4) there is a pause after the swing is complete for the measured values to be displayed.
In the embodiment of
All of the above checks in the compensation analytics engine can be performed in real-time, as the compensation rolling motion of the vehicle is performed. It is not required to gather a complete set of target poses along the full range of the compensation roll to perform the various integrity checks.
When an aligner is in need of repair, a service technician is sent to the site to work on the aligner. Sometimes the issue could be resolved without a costly service call if the service technician could operate the aligner remotely. Methods exist for taking control of an aligner remotely using special software and an internet connection, but they are limited in their capabilities. See, e.g., U.S. Pat. Nos. 8,073,586 and 8,452,484, which are hereby incorporated by reference in their entirety herein.
There is a need for a remote display/interface for the technician on site using the equipment, and a remote display/interface that service personnel at a different location can use to control the aligner, all on a simple readily available device. A solution that easily does both of these in one simple architecture will now be disclosed. In this embodiment, a wireless network is employed to physically disconnect the constituent hardware components of the aligner to each other and to the internet. This enables a wireless device such as a laptop, tablet, mobile phone, or any other device with a web browser to connect to the aligner to provide a remote display/interface for the aligner.
As shown in
In certain embodiments, a wireless network is employed to physically disconnect the constituent hardware components of the aligner. It is desirable that the main processing component of the system which computes the measurement angles remains fixed in front of the vehicle for the duration of the alignment procedure. This component establishes a secure wireless access point that can be accessed by any number of wireless devices. An application server runs software services that support client connections via any standards-compliant web browser. The user can connect to the wireless access point and interact with the browser with a laptop, tablet, mobile phone, or any other device with a web browser. It becomes unnecessary for the technician to physically interact with the fixed measuring device during an alignment procedure. The technician is able to perform all the necessary steps to complete the vehicle alignment in the process of physically adjusting the suspension. This provides both a time and cost savings over current solutions.
Additionally, the disclosed networked aligner can establish a client connection to an external WiFi access point, to provide internet access to the software components running in the system. This is advantageous for software/firmware version management, maintenance, and trouble shooting. Providing the ability to access the equipment over the internet translates directly to more up-time. It becomes unnecessary for a service technician to visit the aligner in person to diagnose issues or to install software updates. This saves the owner time because the machine is typically down waiting for the service technician, which could take hours or days.
The summary and abstract sections may set forth one or more but not all exemplary embodiments of the present invention as contemplated by the inventor(s), and thus, are not intended to limit the present invention and the appended claims in any way.
Embodiments of the present invention have been described above with the aid of functional building blocks illustrating the implementation of specified functions and relationships thereof. The boundaries of these functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternate boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed.
The foregoing description of the specific embodiments will so fully reveal the general nature of the invention that others can, by applying knowledge within the skill of the art, readily modify and/or adapt for various applications such specific embodiments, without undue experimentation, without departing from the general concept of the present invention. Therefore, such adaptations and modifications are intended to be within the meaning and range of equivalents of the disclosed embodiments, based on the teaching and guidance presented herein. It is to be understood that the phraseology or terminology herein is for the purpose of description and not of limitation, such that the terminology or phraseology of the present specification is to be interpreted by the skilled artisan in light of the teachings and guidance.
The breadth and scope of the present invention should not be limited by any of the above-described exemplary embodiments.
Exemplary embodiments of the present invention have been presented. The invention is not limited to these examples. These examples are presented herein for purposes of illustration, and not limitation. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the invention.