Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

The Framework of the Virtual Laser Tracker – A

Systematic Approach to the Assessment of Error sources


and Uncertainty in Laser Tracker Measurement

Dehong Huo, Paul G. Maropoulos, and Chun Hung Cheng

The University of Bath


d.huo@bath.ac.uk, p.g.maropoulos@bath.ac.uk,
c.h.cheng@bath.ac.uk

Abstract. Laser trackers have been widely used in many industries to meet
increasingly high accuracy requirements. In laser tracker measurement, it is
complex and difficult to perform an accurate error analysis and uncertainty
evaluation. This paper firstly reviews the working principle of single beam laser
trackers and state-of-the-art of key technologies from both industrial and aca-
demic efforts, followed by a comprehensive analysis of uncertainty sources. A
generic laser tracker modelling method is formulated and the framework of the
virtual tracker is proposed. The VLS can be used for measurement planning,
measurement accuracy optimization and uncertainty evaluation. The completed
virtual laser tracking system should take all the uncertainty sources affecting
coordinate measurement into consideration and establish an uncertainty model
which will behave in an identical way to the real system.

Keywords: Virtual Laser Tracker, Measurement Uncertainty; Error Modeling;


Simulation, Large Scale Metrology.

1 Introduction
Conventional coordinate measuring machines (CMMs), as one of the powerful meas-
uring instruments, have been widely used for dimensional and geometrical inspection
in manufacturing industry for the past few decades. However, conventional CMMs
are not appropriate when measuring large components (typical size of 5m to 100m).
In addition, in some cases, it is not possible or necessary to bring the parts onto the
CMMs. Therefore, mobile large scale metrological instruments are being used to meet
these requirements. Among a broad range of large scale measuring systems, such as
optical scanner, laser radar, indoor GPS, digital photogrametry, laser trackers have
become the backbone for accurate dimensional measurement in many industrial and
scientific fields due to their high accuracy, large measuring range, high sampling rate
and automatic target tracking, etc (Peggs et al, 2009).
Since introduced in the late 1980s (Lau, 1985), laser trackers have been used
widely in various industries where large scale measurement is increasingly in de-
mand, such as aerospace, automotive, shipbuilding, nuclear, mechanical manufactur-
ing. Specific applications include:

G. Huang et al. (Eds.): DET2009 Proceedings, AISC 66, pp. 507–523.


springerlink.com © Springer-Verlag Berlin Heidelberg 2010
508 D. Huo, P.G. Maropoulos, and C.H. Cheng

• Direct measurement and compensation of multi-axis robot and dynamic robot cali-
bration
• Measurement / Inspection of large optical surface
• Optical system precision alignment
• Aircraft manufacturing and assembly
• Reverse engineering
• Calibration of CMMs and machine tools
• Metrology enabled assembly and automation
ISO 9000:2000 requires that measurement related to the product quality has to be
traceable and a measurement uncertainty according to Standard, e.g. GUM, has to be
set up. In laser tracker measurement it is a complex task to determine uncertainty
sources that can occur during a measurement process and to estimate their contribu-
tion to the measurement uncertainty. Further, ISO 14253-1 ‘GPS - Inspection by
measurement of workpieces and measuring equipments, Part 1: Decision rules for
proving conformance or non-conformance with specifications’ has explicitly included
measurement uncertainty in proving conformance of products (ISO 1998). With the
decrease of product tolerances, accurate assessment of uncertainty therefore has con-
siderable economic impact.
The measurement uncertainty arises from a number of different contributors that
are typically unknown and also most likely to change with time. The effect of a cer-
tain uncertainty sources may vary, depending on a specific measurement task. The
estimation of the effect of a certain uncertainty source is rather complex in real meas-
urement since a single error component generally cannot be separated from others.
Further, the measurement uncertainty of laser trackers is highly non-uniform in space.
This inherent non-uniform distribution of uncertainty in 3D space is caused by the
relative lower accuracy angular measurement and imperfection of the optical compo-
nents. The accuracy specification from manufacturers (e.g. obtained by performing
ASME B89.4.19 conformance tests (ASME 2006)) can generally not describe the
distribution of uncertainty in 3D space. Moreover, the laser tracker operator does not
have a standard measurement procedure to a specific measurement task, therefore
measurement sequence and spatial distribution of the measured point can be different
even for the same part. Uncertainty evaluation of a specific measurement task can be
extremely difficult.
A promising approach to a systematic investigation of uncertainty sources and their
effect in a laser tracker measurement is the simulation of laser tracker and its measur-
ing process in a virtual environment. The virtual laser tracker (VLT) allows a realistic
reproduction of any physical laser trackers in a computer and enables simulation of
laser tracker measurement under accurately specified conditions, thus helping to pro-
vide very detailed information on the measurement uncertainty.
This paper is organized as follows: Section 2 reviews the working principle of sin-
gle beam laser trackers and development of key technologies from both industrial and
academic efforts. Section 3 provides a comprehensive analysis of uncertainty sources
in laser tracker measurement. Section 4 introduces a generic kinematic modelling
approach for a laser tracker Section 5 reviews the uncertainty evaluation methods and
proposes the framework of the virtual laser tracker. Finally, conclusions and future
work are addressed in Section 6.
The Framework of the Virtual Laser Tracker 509

2 State-of-the-Art of Laser Trackers


2.1 Working Principle
A laser tracker can be regarded as a portable frameless coordinate measuring machine
that reports 3D coordinates in a spherical coordinate system. Here the working princi-
ple of a typical single beam laser tracking system is described. The system includes a
laser interferometer, a tracking mirror mechanism, a position detector, a motor control
unit, and a reflector, as shown in Figure-1.

Retro-reflector

Polarizing Beam splitter


beam splitter Tracking mirror
Laser Horizontal
source axis motor
S

2-axis rotary
S encoders
vertical
axis motor

Motor θm, φm
Interferometer Control
Unit Pm=(ρm, θm, φm)T
Position sensitive
detector (PSD)
ρm

Fig. 1. The working principle of the single beam laser tracker with steering mirror

The measuring laser beam is emitted from the laser source, and a portion of the
measuring laser beam which passes through the beam splitter is directed to the retro-
reflector by a 2-axis tracking mirror mechanism. An interferometer measures the lin-
ear displacement of a target retro-reflector. A portion of the return beam is directed by
a beam splitter onto a 2 dimensional optical position detector (for instance, position
sensitive detector, PSD) that senses lateral motion of the target reflector. The resultant
error signal is used by motor control unit under a certain control algorithm to drive the
tracking mirror mechanism so that the displacement measuring beam remains centred
on the target as it moves through the space. 2-axis high-resolution angle encoders
provide the horizontal and vertical angles (θ, φ) of a spherical coordinate system. The
displacement interferometer provides the radial coordinate ρ of the target centre.
When the initial optical path between the centre of the mirror and the home position
(brid bath) is calibrated, the spherical coordinate, Pm=(ρm θm, φm)T, of the reflector can
be obtained in real time from the interferometer and encoder readings. The single
beam laser tracker with tracking mirror based on the working principle in Figure-1
will be used as the example throughout the paper, although various optical and me-
chanical arrangements are being used for commercial and research laser trackers as
will be discussed in the next section.

2.2 State-of-the-Art
There are three major manufacturers of commercial laser trackers: Automated Preci-
sion Inc. (API), Faro Technologies Inc., and Leica Geosystems. All manufacturers
510 D. Huo, P.G. Maropoulos, and C.H. Cheng

produce laser trackers that are capable of measuring in both interferometer and ADM
modes. Each commercial laser tracker mode varies slightly in terms of size, measur-
ing range, resolution and accuracy. Ref (Burge, 2007) provides a detailed comparison
of commercial laser trackers available.
Laser trackers are relatively new measuring instruments, in addition to industrial
development, there is some research focusing on performance improvements of laser
trackers. These research efforts can be divided into three aspects, i.e. development of
laser beam steering mechanism, novel calibration method, and servo design im-
provement. Development on laser tracker multilateration has not been discussed here.

2.2.1 Laser Beam Steering Mechanism


Axis misalignment and manufacturing errors in tracking mirror mechanism introduce
significant measurement uncertainties. In recent years, efforts have been made to im-
prove the beam steering mechanism. Nakamura (2006) developed a beam steering
mechanism using a precision sphere as its bearing. NPL further improved this method
and designed a novel laser tracker for the national standard CMM as shown in
Figure-2(a). In NPL design, a carriage carrying a displacement measuring interfer-
ometer is mounted on the sphere and is free to rotate about the centre of the sphere,
which reduces the axis alignment problems. Virtual centre of the sphere is used as the
reference for the interferometer, providing a further reduction in uncertainty due to
beam steering (Hughes et al, 2000). PTB and NPL jointed developed a new type of

(a) (b)

(c) (d)
Fig. 2. Non-commercial laser tracking system
The Framework of the Virtual Laser Tracker 511

laser tracker named LaserTRACER as shown in Figure-2(d), in which the tracking


mechanism is fundamentally different from that in conventional laser tracker. The
interferometer moves in a gimbal mount around a fixed precision sphere which serves
as a reference mirror to reflect laser beam, and radial and lateral deviations of the
mechanical rotation axes do not significantly affect the measurement accuracy
(Schneider, 2004). Osawa et al (2001) developed a compact laser tracker in which an
articulating hemisphere is used as a tracking mirror as shown in Figure-2(b). A flat
mirror is mounted on the articulating hemisphere which sits on three balls as bearings.
The articulating hemisphere driven by an XY stage can rotate in two directions. The
advantage of this design is that the laser tracker is free from any axis misalignment.
But the limitation is that the hemisphere mirror can only steer laser beams about ±30
degrees around two axes. Figure-2(c) shows a laser tracking system for dynamic
measurement applications developed by Shirinzadeh et al (1998). Their contributions
include performance assessment for different types of position sensitive diode (PSD)
and kinematic modelling for laser trackers.

2.2.2 Novel Calibration Method


Measurement accuracy is critically dependent upon instrument calibration. This as-
pect has been, and continues to be, an important research topic within large scale me-
trology community (Peggs 2009). Novel calibration methods for laser trackers are
being developed. The calibration results are used for estimating the error model pa-
rameters, compensating the laser tracker, and evaluating coordinate uncertainty.
ASME B89.4.19 Standard is the first standard for testing and performance evalua-
tion of laser trackers. It formulates some calibration procedures for both user and
manufacturers. The large-scale coordinate metrology group at NIST has developed a
number of laser tracker calibration facilities, such as 60 m laser rail and laser-rail and
carriage system (LARCS), which can be used for complete set of B89.4.19 tests
(Sawyer, 2003).
However, common tests formulated by ASME B89.4.19 Standard are limited and
may not be enough to extract all individual error parameters (as will be discussed in
the next section). Novel calibration methods are still much needed to make use of
accuracy of laser trackers. Ouyang et al (2005) proposed a laser tracker calibration
method using coordinate measuring machines. Using this method, a commercial laser
tracker was calibrated and angular errors were found to be the key error source. A
rotary guiding system to measure angular errors had been developed as shown in
Figure-3(a) and improvement on measuring accuracy had been reported. Gassner and

(a) (b)
Fig. 3. Examples of calibration methods
512 D. Huo, P.G. Maropoulos, and C.H. Cheng

Ruland (2008) utilized an ultraprecision rotary table to test the performance of the
angular encoder systems in a commercial laser tracker. Figure-3(b) shows the sche-
matic of the technique. A mirror was used in their test instead of retroreflector to
eliminate the effects of an offset between the rotation axes. The encoder angle errors
were significantly improved by applying the angle errors map obtained from the test.
Zobrist et al (2008) developed a system that coupled a commercial laser tracker
with an advanced calibration technique and a system of external stable references.
Their improvement on measurement accuracy lies in a calibration in a geometry that
matches the measurement geometry although this approach is normally difficult or
impossible in many cases.

2.2.3 Servo Design Improvement


Conventional approach using PID algorithm cannot meet the requirements of high
accuracy and fast target trajectory tracking, many investigations on the servo design
for advanced laser tracker control have been conducted. Yen et al (1996) included the
relationship between the tracking angle rotations and measured laser beam offsets in
the control system model and applied H∞ optimization technique for the controller
synthesis. High speed and high accuracy laser tracker measurement was achieved by
compensating the high nonlinear errors and maintaining the laser beams close to the
measurement points. Shi et al (2005) proposed an adaptive PID controller based on
artificial neural network in laser tracker control system to deal with the control system
nonlinear behaviour.

3 Sources of Uncertainty in Laser Tracker Measurement

3.1 Classification of Uncertainty Source

A laser tracker is an integrated measuring system with optical, mechanical, and elec-
tronic components. The errors from those components, together with the environ-
mental and operational factors, determine the measurement uncertainty of the LTS.
There are few literatures systematically addressing classification of uncertainty
sources and their effects on laser tracker measurement.
Uncertainty sources can be categorized in a number of different ways. For instance,
Wilhelm divided uncertainty into five main categories: hardware, workpiece, sam-
pling strategy, fitting and evaluation algorithm, and extrinsic factors (Wilhelm et al,
2001). Trapet and Waldele (1996) have categorized uncertainty sources into two
groups: those that are accessed by measurements and those that are normally esti-
mated. There are also some other classification methods, although these classifica-
tions were normally used for coordinate measuring machines with contacting probes.
This paper classifies uncertainty sources in laser tracker measurement into four
categories: namely, static or quasi-static uncertainty sources, dynamic uncertainty
sources, fitting and evaluation algorithm related uncertainty sources, and measuring
strategy/sequence related uncertainty sources. An overview of the uncertainty sources
in a laser tracker measurement is given in Figure-4.
The Framework of the Virtual Laser Tracker 513

Geometric errors: Dynamic errors:


ƒ Manufacturing errors Static/quasi-static Dynamic ƒ instrument vibrations
ƒ Assembly uncertainty source uncertainty source ƒ workpiece vibrations
misalignment ƒ servo errors from
ƒ Laser beam offset mirror steering control
ƒ Imperfection of the system
SMRs ƒ Acceleration of
ƒ Datum point errors Total Uncertainty in reflector
Laser Tracker
Measurement Measuring strategy:
Non-geometric errors: ƒ Measuring sequence
ƒ refractive index of air ƒ Position of laser
ƒ thermal induced trackers
distortions ƒ Orientation of laser
ƒ thermal induced laser
Fitting and Measuring strategy
trackers
source variation evaluation algorithm and sequence ƒ Scanning rates

Fig. 4. Overview of the uncertainty sources in laser tracker measurement

Static or quasi-static uncertainty sources arise from those errors that are slowly
varying with time and related to the structure of the laser tracker itself and laser wave-
length. These sources can be subdivided into geometry uncertainty sources and non-
geometry sources. Geometric uncertainty sources include mechanical inaccuracies of
the rotational axes misalignment, mirror centre offset in a single-beam laser tracking
system, spherically mounted retroreflector (SMR) imperfections, and datum point
errors (bird bath distance), etc. Those uncertainty sources are related to systematic
errors which normally account for the majority of the measurement uncertainty and
are the major focus of the error correction. All the commercial laser tracker manufac-
turers have their own error correction algorithms and calibration procedure.
Non-geometry sources include laser beam wavelength variation which is due to the
variation of the air refractive index. The well-known Edlen equation and its updated
version describe this effect, i.e. the wavelength is determined by barometric pressure,
temperature, relative humidity and chemical composition. The laser wavelength varia-
tion can cause distance measurement errors on the order of magnitude of a few mi-
crons. Thermal induced distortions of optical and mechanical components and laser
source variation can also be categorized into this group.
Dynamic uncertainty sources include measuring instrument vibrations, workpiece
vibrations, acceleration of target when using scanning mode, and servo errors from
mirror steering control system (both static and dynamic following errors) etc. Some of
them are considered to be relatively negligible and difficult to evaluate. But some
dynamic sources such as servo errors are deterministic and need to be well assessed.
A real laser tracker measurement may involve complex shape with tight tolerance
and thousands of points to be measured. Because the measurement uncertainty of la-
ser tracker is highly non-uniform in space, measuring strategy and sequence have sig-
nificant influence on the uncertainty. Further, fitting and evaluation algorithm for the
discrete measured points would also introduce uncertainty to the measurement results
and cannot be neglected. An analysis of uncertainty sources caused by measurement
sequence, position and orientation of the laser tracker, and fitting and evaluation algo-
rithm becomes important to the complex laser tracker measurements.
It should be noted that no matter which classification scheme is chosen, the most
important thing is to determine all the sources of uncertainty either quantitatively or
statistically. Hence an accurate and reliable uncertainty evaluation can be performed.
514 D. Huo, P.G. Maropoulos, and C.H. Cheng

Table 1. Kinematic error sources in laser tracker measurement

Error parameters Definition


Laser Rotational Mirror plane The distance between actual mirror plane and the
tracker mirror offset, tmv rotation centre
head and plane Mirror tilt error, The angle between the actual mirror plane and the
IFM rmv vertical circle axis
Laser beam Laser beam The distance of actual laser beam to axis v and w
orientation offset, tbw and tbv are defined as tbw and tbv respectively
Laser beam tilt, The angles between the actual laser beam and axis
rbh and rbv h and v are defined as rbh and rbv respectively.
Axis Axis alignment Due to manufacturing and assembly errors, two
alignment offset, thv axes do not intersect each other. The distance
between two axes is defined as axis alignment offset
Axis alignment Ideally two axes should be perpendicular to each
tilt, rhv other. The deviation angle of actual axis v to its
ideal position
Encoders Encoder centre The distance between the encoder centre and
offset, eh and ev mirror rotation centre
SMR Reflector Vertex centering The deviation of actual SMR vertex to the ideal
offset sphere centre, both radial and lateral directions
Dihedral angle In ideal cube corner the angle between each of the
three pairs of mirror faces is exactly 90˚. The angle
difference to the ideal is called dihedral angle error
Notation:
h – horizontal circle axis
v – vertial circle axis
w – a virtual axis perpendicular to h and v, and also lie through the intersect point of the h and v
t – misalignment translational errors or offset errors
r – misalignment rotational errors or tilt errors

3.2 Independent Kinematic Parameters


The independent kinematic parameters describe physical position and orientation
errors of the optical and mechanical components. The kinematic model must be com-
plete, that means it must cover all independent geometric error sources and no redun-
dant parameters (Zhuang and Roth, 1995). For the laser tracker with a beam steering
mirror and standing IFM emitting laser beam as shown in Figure-5, kinematic error
sources have been identified and categorized. Eight kinematic independent errors are
proposed and used to describe the mirror, rotating axes and laser beam misalignment
error inside the tracker. Three independent errors are used to examine the influence
from imperfection of the retro reflectors as described in ASME B89.4.19. Together
with rotary encoder errors, a comprehensive kinematic error model can be established
for this type of laser trackers. Table-1 summarizes these error sources in the laser
tracker measurement and their definitions are explained.
It should be noted that individual laser trackers and reflectors may have different
optical and mechanical component arrangement, the independent errors components
discussed above may not apply to all the laser trackers and reflectors, but the method
can be referred to as a guide to identify the uncertainty sources of such instruments.
The Framework of the Virtual Laser Tracker 515

With the ideal kinematic model of laser trackers and the independent geometric er-
rors, one can perform a comprehensive error sensitivity analysis to understand how a
particular error propagates to the overall uncertainty through the simulation. Zhuang
and Roth (1995), Lin and Lu (2005) and Loser and Kyle (1999) have done some work
on the error modelling and sensitivity analysis respectively.

4 Kinematic Model of Laser Tracker with Beam Steering Mirror


Establishing an accurate kinematic model for laser trackers is essential for model-
based calibration and uncertainty evaluation. It can also be used for servo control of
the laser trackers (Zhuang, 1995). Accurately modelling the laser tracker will be the
core part of the virtual laser tracker proposed in this paper. However the manufactur-
ers keep the actual models confidential and little information has been published so
far, which makes it difficult for researchers to advance the laser tracker development
and make use of the laser tracker.

4.1 Ideal Kinematic Model


This section deals with the kinematic model of a laser tracker with the beam steering
mirror. Since the actual laser trackers vary in optical and mechanical arrangement as
discussed in section 2.2, the kinematic model here obviously cannot be applicable to
all the laser trackers. However it aims to provide a generic modelling method to per-
form geometric error analysis. To establish a kinematic model of the single beam la-
ser tracker, the following four assumptions are made for an ideal laser tracker:
• Two rotary axes are intersecting and perpendicular, i.e. there are no axis misalign-
ment errors.
• One rotary axis (normally the horizontal axis, referring to Figure-5) lies on the mir-
ror centre.

Zb Rotary axis 1
(vertical) Xm
φ

Yb, Ym
Rotary axis 2
(horizontal) θ

Xb O
P
er
eo Laser
source
ei
θ0 Zm
Laser source

Fig. 5. The kinematic model of the laser tracker steering mirror (other optics such as splitters
are neglected for simplicity)
516 D. Huo, P.G. Maropoulos, and C.H. Cheng

• The incident laser beam hits the mirror at its centre.


• The incident laser beam is parallel to one rotary axis (normally the vertical axis,
referring to Figure-5).
Two Cartesian coordinate systems (CS) – the base CS (Xb, Yb, Zb) and the mirror CS
(Xm, Ym, Zm) are defined as shown in Figure-5, the origins of the two CS are placed at
the mirror centre O. Zb is the rotation axis of the vertical axis, and Ym is the rotation
axis of the horizontal axis. Zm is normal to the mirror surface. The base CS (Xb, Yb, Zb)
is the fixed global coordinate frame with reference to the laser tracker stationary
body. When the mirror is at its home position (φ = θ = 0), Ym and Yb are coincident,
and there is an initial angle θ0 between Xm and Zb. the incident laser beam is defined to
be parallel to Zb, The above assumptions are made without loss of generality. When
the mirror is at an arbitrary position, i.e. with arbitrary rotation angles φ and θ, the
following homogeneous transformation matrix (HTM) are used to define the position
and orientation of the mirror CS with respect to the base CS.
b
Tm= Rot(z, φ)Rot(y, θ)Rot(x, θ0)Rot(x, 90o) (1)
Let the incident beam be Li and its unit direction vector be ei, the reflected beam be Lr
and its unit direction vector be er, and the mirror normal direction and unit direction
vectors be Lo and eo respectively.
The unit direction vector eo with respect to mirror CS is meo = [0 0 1]T. Transform it
to the base CS, beo is obtained:
b
eo = bTm meo (2)
The direction of laser beam vector, Li, is fixed. The mirror surface normal unit vector
eo can be obtained by Equation-2. Ray tracing analysis can be performed to derive the
reflected beam which varies with the position of the target. According to the ray trac-
ing method, suppose an incident laser beam starts at point S = (sx sy sz 1)T with unit
direction vector ei = (eix eiy eiz)T. The laser beam strikes the centre of the mirror, O,
and the mirror surface normal is eo = [eox eoy eoz]T, then it is reflected to the target point
P = (px py pz 1)T with unit direction vector er = [erx ery erz]T. P and er are obtained by:
P = R(eo) S (3)
er = r (eo) ei (4)
where R(eo) and r (eo) are transformation matrix with respect to the mirror normal.
R(eo) =
⎛1 − eox eix / ρ − eoy eix / ρ − eoz eix / ρ deix / ρ ⎞
⎜ ⎟
⎜ − eox eiy / ρ 1 − eoy eiy / ρ − eoz eiz / ρ deiy / ρ ⎟
⎜ −e e /ρ − eoy eiz / ρ 1 − eoz eiz / ρ deiz / ρ ⎟
(5)
⎜ ox iz

⎜ 1 ⎟⎠
⎝ 0 0 0

⎛ 1 − 2eox2 − 2eox eoy − 2eox eoz ⎞


⎜ ⎟
r (eo) = ⎜ − 2eox eoy 1 − 2e − 2eoy eoz ⎟
2
oy (6)
⎜ − 2e e − 2eoy eoz 1 − 2eoz2 ⎟⎠
⎝ ox oz
The Framework of the Virtual Laser Tracker 517

where d is the distance from the coordinate origin to the mirror surface, and ρ is
given by
ρ = eo· ei (7)
The target position can also be written as
P = lr · er (8)
Where lr is the amplitude of the vector Lr, i.e. the distance from the point where the
laser strikes the mirror to the target. Substitute Equation-4 and -6 to Equation-8, the
target position, P, is given by

P = lr · r (eo) ei (9)
In the laser tracker measurement, lr is calculated by the interferometer, eo can be com-
puted by 2-axis encoder readings, so the arbitrary target position in ideal condition
can be obtained from Equation-1 to 9.

4.2 Modelling Kinematic Errors

As an example, measurement errors due to mirror plane offset errors are analyzed in
this section. Other kinematic errors can be modelled in a similar way. Ideal mirror
plane should be on the rotation centre where two axes intersect. The distance between
actual mirror plane and the rotation centre is defined as tmv. When there is a mirror
plane offset error tmv, the mirror plane lies in the tangential plane of a circle with the
centre of rotation axis and a radius of tmv as shown in Figure-6. When the mirror is at
its initial position with an initial angle θ0 between the incident beam and the mirror
plane, the incident beam hit the mirror at point O0. When mirror rotates to angle θi to
track the target point Pi, the incident beam hits the mirror at point Oi. Mirror plane
offset errors result in position change of the intersection point and hence introduce
measurement errors. Assume the interferometer reading at initial position be zero, and
let the measured coordinate at Pi be Pmi = (ρmi, θmi, φmi)T. The range measurement er-
ror, Δρi, due to mirror plane offset can be obtained by

Δρi = O0 Pi − ρ mi (10)

where O0 Pi denotes the length of the vector O0Pi, same definition for others. ρmi can
be written as

ρmi = Oi S + Oi Pi − O0 S = Oi Pi − O0 Oi (11)

The intersection point change O0 Oi can be obtained from Figure-6.


t mv t
O0 Oi = − mv (12)
sin θ 0 sin θ mi
518 D. Huo, P.G. Maropoulos, and C.H. Cheng

Actual mirror position


Corrected mirror position
Initial mirror position

Pi
O
tmv
Lr,i

O0
L r,0 P0
θ i’
θ0 O
i
Lo,0
c Lo,i’
Lo,i
θi
Li

Fig. 6. Modelling of mirror plane offset errors

O0 Pi can be obtained according to the geometric relationship in Figure-6.


O0 Pi =
((O O )
0 i
2
+ (Oi Pi ) 2 − 2O0 Oi ⋅ Oi Pi cos 2θ mi )
1/ 2
(13)

Substitute Equation-11 to -13 into -10 the range measurement error, Δρi, can be de-
termined.
Let Δθi be the measurement error of angle
θmi. Its value can be obtained from Figure-6. using a similar way.

⎛ Oi Pi ⎞
Δθi = θ m i + 12 arcsin⎜⎜ sin 2θ mi ⎟⎟ − 90° (14)
⎝ O0 Pi ⎠
Where Oi Pi and O0 Pi can be calculated from Equation-11 to -13.

5 Framework of the Virtual Laser Tracker

5.1 Uncertainty Evaluation Methods

The Guide to the Expression of Uncertainty in Measurement, GUM (BIPM et al,


1995) is the internationally accepted document for uncertainty evaluation. As speci-
fied in the GUM, one must first indentify each uncertainty source, quantify its stan-
dard uncertainty, and determine its sensitivity coefficient and correlation with other
uncertainty sources. The product of each standard uncertainty by its sensitivity coeffi-
cient is then added in quadrature to obtain the combined uncertainty. Finally the
The Framework of the Virtual Laser Tracker 519

combined uncertainty is multiplied by an appropriate coverage factor to yield an ex-


panded uncertainty.
The GUM method is extremely general and only useful when a well-defined
mathematical model of the measurement process can be ascertained (Wilhelm et al,
2001). For real coordinate measurement, e.g. laser tracker measurement, the meas-
urement model is too complicated to obtain derivatives for sensitivity coefficient of
each uncertainty source, the GUM method is usually not applicable. In recent years, a
number of uncertainty evaluation approaches based on numerical simulation, mainly
the Monte Carlo method, have been developed. Supplement 1 to the GUM formulates
the simulation method (BIPM et al, 2006) and recommends a Monte Carlo method
(MCM) as a basic tool for this approach.

5.2 Framework of the Virtual Laser Tracker (VLT)

The concept of virtual coordinate measuring machine (Virtual CMM) was first pro-
posed by PTB in 1996 (Trapet and Waldele, 1996), NPL and NIST have done some
further research in the area of Virtual CMM respectively. However, the research on
Virtual CMM or evaluation of measurement uncertainty based on computer simula-
tion currently mainly focus on contact probe-based conventional CMM, to the best of
our knowledge, there are no publications available on virtual laser trackers. Laser
trackers can be viewed as 3D coordinate measuring machines and share many com-
mon attributes with conventional Cartesian CMMs. However, there are also signifi-
cant differences between the laser tracking system and conventional CMMs in terms
of kinematics, uncertainty distribution and sources, etc., which make the modelling of
laser tracking system and its measurement process difficult.
The paper proposes the framework of the virtual laser tracker (VLT). VLT is in es-
sence a computer description of the real laser tracker simulating a laser tracker’s op-
eration and its measurement process in a virtual environment. It also enables laser
tracker off-line measurement to take place exactly as if an operator operates the laser
tracker and moves a real reflector on the part to be measured. Ideally a VLT com-
prises of the accurate model of real laser trackers and should hence be identical to
the simulated real laser tracker. All the kinematics, dynamics and uncertainty sources
must be covered and modelled based on the actual behaviour of the real laser trackers.
Figure-7 shows a schematic of the framework of the proposed VLT. Given a spe-
cific measurement task, a conventional laser tracker measurement involves taking real
measurement for N points on the workpiece under designed strategy and sequence
against CAD models. The measured raw data will be inputed directly to the laser
tracker analysis software for computation of the geometric parameters. Integrated
with the calibration report or manufacturers’ specifications, a measurement result is
generated as an estimate of the measured geometry. This conventional approach is
straightforward but fails to make most use of the laser tracker accuracy to meet in-
creasingly tight tolerances, further, measurement planning and optimization can only
be performed by trial and error.
When integrating with VLT, the laser tracker measurement can be conducted either
online or offline (i.e. independent of the real measurement). When online operating
VLT the real measured raw data are inputed into the uncertainty evaluation engine of
the VLT as initial points to start the simulation (such as Monte Carlo simulation) from
520 D. Huo, P.G. Maropoulos, and C.H. Cheng

which a better estimate of uncertainty will be generated based on statistical principle


and most importantly the accurate model of the system. Finally the simulated results
rather than raw measurement data are inputed to the laser tracker analysis software to
generate the measurement report. When offline operating VLT CAD models (or a
measurement plan) are brought to VLT, the measurement process is simulated against
the CAD models, and uncertainty can be assessed and optimized.
The proposed VLT framework consists of three key components – a laser tracker
model, a laser tracker simulator and visualization module, and an uncertainty evalua-
tion engine. Error analysis, uncertainty evaluation and measurement process optimiza-
tion are implemented based on the three key components.

Identification of LTS calibration assessment General measurement


Uncertainty sources B89.4.19 or other tests uncertainty
Individual error
Modelling of optical parameters
elements and system
Virtual Laser Tracker
Model of LTS including all LTS simulator and
major uncertainty sources visualization module

Kinematic Other Uncertainty Point coordinate


model of LTS models evaluation engine uncertainty
(GUM / Monte Carlo)

CAD models

Report
Specific measurement Real measured General LTS Measurement results with
task and strategy points analysis software uncertainty for specific task
Conventional LTS measurements

Integration with Virtual LTS

Fig. 7. The framework of the VLT

• Laser tracker models. The effective operation of VLT relies on an accurate mod-
elling laser tracker system. A comprehensive model of the laser tracker should in-
clude detailed description of the system, all the major uncertainty sources, and their
propagation effects to coordinate measurements. The laser tracker model can be
divided into several sub-models, e.g. the models describing kinematics of optical
and mechanical parts, servo control system, thermal-induced uncertainty source,
and vibration and dynamic effects, etc.
• Laser tracker simulator and visualization module. Simulator and visualization
module is used to provide solutions for error analysis in algebraic form and visual-
izes simulation results. Although there are a number of optical design software
packages available for ray tracing, they generally cannot provide solutions in
algebraic form and hence cannot accurately model the laser trackers. Therefore at
this stage in-house software is being developed. By running the VLT simulator
and visualization module one can visually check how each uncertainty source af-
fects the measured point coordinates uncertainty against a measurement plan for a
specific task. Hence it is possible to make use of this advantage for the measure-
ment planning and measurement accuracy optimization. Simulator and visualiza-
tion module also facilitates and enhances the calibration process.
The Framework of the Virtual Laser Tracker 521

• Uncertainty evaluation engine. Error parameters and their associated uncertainty


are used as input for the uncertainty evaluation engine from which the combined
coordinate uncertainty can be calculated. The Monte Carlo method or other simula-
tion based method is used for propagating input parameters.
The VLT is a platform for measurement planning and measurement accuracy optimi-
zation. By integrating laser tracker measurement with the VLT, error analysis,
uncertainty evaluation and measurement process optimization are performed in a sys-
tematic and predictable manner. The completed virtual laser tracking system should
take all the uncertainty sources affecting coordinate measurement into consideration,
and establish an uncertainty model which will behave in an identical way as the real
system. In order to obtain accurate uncertainty evaluation in the whole working vol-
ume, the dedicated calibration from which individual error components can be deter-
mined is critical. Uncertainty of any point in 3D space will be predicted by the model
for a specific measurement task.

6 Conclusions and Future Work


Based on the comprehensive review on the state-of-the-art of laser tracker measure-
ment and uncertainty sources, a generic laser tracker modelling method was formu-
lated. Laser tracker model serves the core part for the proposed framework of the
VLT. The VLS can be used for measurement planning, measurement accuracy opti-
mization and uncertainty evaluation. The completed VLT should take all the uncer-
tainty sources affecting coordinate measurement into consideration and establish an
uncertainty model which will behave in identical way as the real system. Although the
framework of the VLT has been outlined in the paper, the development of VLT is still
at the nascent stage. Future efforts will be directed towards the following aspects:
• Accurate mathematic models for laser trackers are needed for successfully per-
forming laser tracker measurement with VLT. However, Individual error compo-
nents in the model can normally be provided neither by machine acceptance tests
(such as ASME B89.4.19) or interim tests. Unfortunately, little effort has been
made in this area. Therefore, it is imperative to design novel calibration methods
and error models to quantitatively determine individual errors required.
• The evaluation of thermal and dynamic effects including servo control system is
necessarily incorporated into an uncertainty model.
• Efforts will be put on the commercial laser tracker modelling if possible with the
aim of benefiting the wide audiences of the large scale metrology community.

Acknowledgement
The authors wish to acknowledge the financial support of the Engineering and Physi-
cal Sciences Research Council, UK (Grants EP/E002617/1 and EP/E00184X/1) and
the substantial help of our industrial partners.
522 D. Huo, P.G. Maropoulos, and C.H. Cheng

References
B89.4.19, Performance Evaluation of Laser-Based Spherical Coordinate Measurement System.
American Society of Mechanical Engineers (2006)
BIPM, IEC, IFCC, ISO, IUPAC, IUPAP, and OIML, Guide to the expression of uncertainty in
measurement, GUM, ISO, PD 6461-3 (1995)
BIPM, IEC, IFCC, ISO, IUPAC, IUPAP, and OIML, Evaluation of measurement data – Sup-
plement 1 to the Guide to the expression of uncertainty in measurement – Propagation of
distributions using a Monte Carlo method (2006)
Burge, J.H., Su, P., Zhao, C., Zobrist, T.: Use of a Commercial Laser Tracker for Optical
Alignment, Optical System Alignment and Tolerancing. In: Proceeding of SPIE, vol. 6676
(2007)
Gassner, G.L., Ruland, R.E.: Laser Tracker Test Facility at SLAC – Progress Report. In: The
10th International Workshop on Accelerator Alignment, February 11-15 (2008)
Hughes, E.B., Wilson, A., Peggs, G.N.: Design of a High-accuracy CMM based on multi-
lateration techniques. Annals of the CIRP 49(1), 391–394 (2000)
ISO 14253-1 GPS - Inspection by measurement of workpieces and measuring equipments, Part
1: Decision rules for proving conformance or non-conformance with specifications (1998)
Lau, K., Hocken, R., Haynes, L.: Robot performance measurement using automatic tracking
techniques. Robot Computer Integrated Manufacturing 2(3), 227–236 (1985)
Lin, P.D., Lu, C.H.: Modeling and Sensitivity Analysis of Laser Tracking Systems by Skew-
Ray Tracing Method. Journal of Manufacturing Science and Engineering 127, 654–662
(2005)
Loser, R., Kyle, S.: Alignment and field check procedures for the Leica Laser Tracker LTD
500. Boeing Large Scale Optical Metrology Seminar (1999)
Nakamura, O., Goto, M., Toyada, K., Takai, N., Kurosawa, T., Nakamata, T.: A Laser Tracking
Robot-Performance Calibration Syatem using Ball-Seated Bearing Mechanisms and a
Spherically Shaped Cat’s Eye Retroreflector. Review of Scientific Instruments 65(1), 1006–
1011 (1994)
Osawa, S., Tashiyuki, T., Hong, J., Noguchi, H., Kurosawa, T.: Evaluation of the Performance
of a Novel Laser Tracker Used for Coordinate Measurements, Recent Development in
Traceable Measurements. In: Proceedings of the SPIE, vol. 4401, pp. 127–135 (2001)
Ouyang, J., Liu, W.L., Sun, D.X., Yan, Y.G.: Laser Tracker Calibration Using Coordinate
Measuring Machine. In: Proceedings the ASPE, October 9-14 (2005)
Peggs, G.N., Maropoulos, P.G., Hughes, E.B., Forbes, A.B., Robson, S., Ziebart, M., Muralik-
rishnan, B.: Recent developments in large-scale dimensional metrology. Journal of Engi-
neering Manufacture, Proc. IMechE, Part B 223, 1–26 (2009)
Sawyer, D.S., Fronczek Jr, C.: Laser Tracker Compensation Using Displacement Interferome-
try. In: Proceedings the ASPE, October 26-31 (2003)
Schneider, C., Engelhardt, M.: LaserTracer - A New Type of Self Tracking Interferometer. In:
8th International Workshop on Accelerator Alignment, Geneva, Switzerland, 4-7 October
(2004)
Shi, Y., Zhang, G., Li, X.: Design and Improvement of Laser Tracking Control System Using
ANN Technique. Advanced Materials and Devices for Sensing and Imaging II. In: Proceed-
ings of the SPIE, vol. 5633, pp. 556–561 (2005)
Shirinzadeh, B.: Laser-interferometry-based tracking for dynamic measurements. Industrial
Robot 25(1), 35–41 (1998)
Trapet, E., Waldele, F.: The Virtual CMM Concept. In: Advanced Mathematical Tools in Me-
trology II, pp. 238–247. World Scientific, Singapore (1996)
The Framework of the Virtual Laser Tracker 523

Wilhelm, R.G., Hocken, R., Schwenke, H.: Task Specific Uncertainty in Coordinate Measure-
ment. Annals of the CIRP 50(2), 553–563 (2001)
Yen, J.Y., Jeng, C.S., Fan, K.C.: Servo Design for a 3-D Laser Tracking Measurement system.
Journal of Dynamic Systems, Measurement, and Control 118, 476–481 (1996)
Zhuang, H., Roth, Z.S.: Modeling Gimbal Axis Misalignments and Mirror Center Offset in a
Single-Beam Laser Tracking Measurement System. International Journal of Robotics Re-
search 14(3), 211–224 (1995)
Zobrist, T.L., Burge, J.H., Davison, W.B., Martin, H.M.: Measurements of large optical sur-
faces with a laser trackerAdvanced Optical and Mechanical Technologies in Telescopes and
Instrumentation. In: Proceeding of SPIE, vol. 7018 (2008)

You might also like