Home Search Collections Journals About Contact us My IOPscience

A large-scale laser plane calibration system

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2007 Meas. Sci. Technol. 18 1768

(http://iopscience.iop.org/0957-0233/18/6/S16)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 129.93.16.3
This content was downloaded on 06/09/2015 at 08:10

Please note that terms and conditions apply.

IOP PUBLISHING MEASUREMENT SCIENCE AND TECHNOLOGY
Meas. Sci. Technol. 18 (2007) 1768–1772 doi:10.1088/0957-0233/18/6/S16

A large-scale laser plane calibration

system
Liqun Ma1,2, Liding Wang2, Tieze Cao1, Jihu Wang1,
Xiaomei He1 and Changyou Xiong1
1
Beijing Changcheng Institute of Metrology and Measurement, China Aviation Industry
Corporation I, Beijing 100095, People’s Republic of China
2
Dalian University of Technology, Dalian 116024, People’s Republic of China
E-mail: [email protected]

Received 26 September 2006, in final form 6 March 2007

Published 8 May 2007
Online at stacks.iop.org/MST/18/1768

Abstract
A large-scale laser plane calibration system is developed. This plane is
constructed by a 35 m long horizontal granite guide rail and a 1.5 m long
vertical metal guide rail. Two dual-frequency interferometers are used to
measure the coordinates of a target moving on the plane. One of them can
measure the horizontal displacement while eliminating the Abbe errors
caused by the straightness of the granite guide rail. Another measures the
sum of horizontal displacement and vertical displacement. This laser plane
can be used as a lab standard to calibrate portable coordinate measuring
machines by measuring the target on the plane. Calibration tests for a laser
tracker are given.

Keywords: large-scale metrology, laser plane, laser tracker, calibration

(Some figures in this article are in colour only in the electronic version)

1. Introduction of the calibrated length can be adjusted with the supporting

system. Such methods are facilitated to perform a comparison
Nowadays, portable CMMs such as laser trackers are widely between laser trackers. But they cannot express the coordinate
employed in manufacturing industry. They are flexible and measurement uncertainty of the tracker very well. In most
have high accuracy, and are becoming the tool of choice cases, coordinate uncertainty is an important consideration in
for large-scale coordinate measuring needs in industry [1]. measuring control points. No large artefacts have been suitable
The critical requirement is to develop efficient technical for this purpose until now. Here a laser plane as a large fixed
approaches to assess the trackers and ensure the integrity of artefact is presented to calibrate a tracker in the laboratory.
their measurement results [2].
In general, a reference length needs to be created
and compared with laser trackers. The American 2. Requirement for calibrating a laser tracker and
Society of Mechanical Engineers (ASME) issued a standard minimum standard plane
B89.4.19 ‘Evaluation of Laser Based Spherical Coordinate
Measurement Systems’ in 2001. In this standard, requirements A laser tracker is a spherical coordinate measuring system, as
and methods for specifying and testing the performance of a shown in figure 1. In this spherical system, a point P (x, y, z)
laser tracker are established; some test procedures are provided is represented by a distance d and two angles (horizontal angle
for checking angle, range, and volumetric uncertainty of a α and vertical angle β). Coordinates (x, y, z) of point P in a
tracker concerning point-to-point distance measurements. To right-handed rectangular system have the following relations
meet the standard, NIST (National Institute of Standards and with the coordinates (d, α, β) in a spherical system:
Technology) developed a Laser Rail Calibration System for 
calibration of a laser tracker. The instrument consists of a x = d cos β sin α

y = d cos β cos α (1)
guide rail for the laser interferometer and supporting system, 

and provides a standard length at job site. The orientation z = d sin β.

0957-0233/07/061768+05$30.00 © 2007 IOP Publishing Ltd Printed in the UK 1768

 A large-scale laser plane calibration system

P(x, y, z)
Z
d

β
Y
O α

X
Figure 1. A point in a spherica1 coordinate system.

Set md as a distance d standard uncertainty, mα as a

horizontal angle α standard uncertainty and mβ as a vertical
angle β standard uncertainty; also set (ux , uy , uz ) as the
standard uncertainty of point P coordinates (x, y, z). The Figure 2. Photo of the laser plane system.
standard uncertainty up of a point is calculated from expression
(1) using the law of propagation of uncertainty:
1.5 m long vertical metal rail. The vertical rail is mounted
u2p = m2d + d 2 cos2 βm2α + d 2 m2β on a pneumatic working platform which slides along the
(2)
if mα = mβ then u2p = m2d + d 2 (1 + cos2 β)m2β . horizontal rail. A target carrier driven by a servo-motor is
For the purpose of evaluating laser tracker performance set on the vertical rail and can move along it. Two dual-
using a laser standard plane, the plane size should be frequency interferometers are used as the laser plane standard.
considered first. Some factors such as cost, space of lab A pentaprism is also used to keep the two interferometers’ light
and environment control restrict the plane dimensions. It is axes having a right-angle intersection. This is the precondition
unreasonable to establish a plane to fit the largest working to make up a rectangle plane coordinate system. Figure 2 is a
volume of a tracker (generally, a tracker working range is: d: photo of the laser plane system.
30 m, β: ±45◦ , α: ±235◦ ), but a minimum plane is needed The calibration system is established in an air-conditioned
for tracker calibration. This minimum plane should be big standard laboratory with temperature well controlled. All
enough to cover the possible direction changes when using a the laser sources, splitters, benders, interferometric mirrors,
tracker to measure a point which is not far from the tracker. In reference reflectors and receivers are set on a stable platform
equation (2), it is known that the standard uncertainty up is and all the object reflectors on a moving platform, so the
associated with d and md , β, mα and mβ . By setting a tracker temperature effects on all light are the same and the dead-path
away from a target at different distances d, the contribution of lengths of every interferometric beam are equal, regardless of
md to up can be tested, so the problem is how to test mα and mβ the vertical light path. To make a reasonable compensation for
within the working range of angles. If a tracker is set up at a the temperature effect on 35 m long light paths, 16 sensors are
distance 1.5 m away from the plane, the plane should be 1.5 m set above the horizontal granite rail, and there is a 2 m interval
high to meet a target vertical angle β changing from 0◦ to 45◦
between sensors. The environment compensation should be
(or −45◦ to 0◦ ). To avoid an obvious change in the distance
done by selecting the temperature sensors corresponding to
between the target moving on the plane and the tracker, the
the light paths.
width of the plane should be limited; it should cover the
angle α changing from 0◦ to 90◦ . In such a case, the ratio
of the shortest distance between the target and the tracker 4. Optic structure of the plane
to the longest distance is about 0.7. So the plane will be a
minimum 4.2 m wide and 1.5 m high when a tracker is set up 4.1. Design for an optic structure
at a distance 1.5 m.
A laser plane with two axes (horizontal axis and To facilitate performance comparisons for angle, range and
vertical axis) can not only verify the coordinate measurement coordinate uncertainty, the optic structure of the plane is
uncertainty of a tracker but also meet the standard B89.4.19. designed in the way shown in figure 3.
The 35 m long horizontal axis of the plane can give a set In figure 3, laser 1 is the standard of the horizontal
of reference lengths for distance d tests of a tracker and just axis of the laser plane; light from laser 1 is separated into
keep target positioning in approximately the same orientation three parts, which run parallel to the horizontal granite rail
relative to the measurement beam. Also, the horizontal axis to their retroreflectors on the moving platform. These three
of the plane can be used for angle α tests and the vertical axis path lights measure the same movement of the platform. As
of the plane for angle β tests of a tracker. the set-up positions of their retroreflectors are different, their
travel lengths are not equal when the platform has deflection
3. Configuration of the system introduced by the straightness of the horizontal granite rail. If
the positions of the retroreflectors are known, the deflection
The laser plane in this project consists of two rails: one is a angles of the moving platform in two directions will be known
35 m long horizontal granite guide rail and the other is a by using the deviations of the three travel lengths. With the

1769
L Ma et al

Figure 3. Optical structure of the coordinate plane: (1) splitter, (2) bender, (3) interferometer, (4) retroreflector, (5) receiver, (6) pentaprism.

known deflection angles, the moving length of any position in

the platform can be calculated. P (x, y, z)
Laser 2 is the standard of 2D plane coordinates. Its beam M4 M3
travels in the same direction as laser 1 to a pentaprism which OZ
Z
is set on the moving platform, which then turns the direction
of the beam at a 90◦ angle to its retroreflector mounted on the
target carrier. This bent beam is the main body of the plane; X
its path length will be changed when the horizontal platform M1
OX
or the target carrier moves. As laser 1 can give the moving
OXZ
distance of the pentaprism mounted on the moving platform,
Y N (x, y, 0) M2 OY
so the movement of the target carrier can be known.
As laser 1 uses three interferometric beams to measure
the displacement of the platform, so it can not only eliminate
the Abbe errors which are introduced by the straightness of the
Figure 4. Arrangement of retroreflectors and pentaprism. M1 , M2 ,
horizontal guide rail and calculate the horizontal movement M3 and M4 : retroreflectors; N: pentaprism.
of a target on the plane accurately but also measure the rail
straightness and the offsets of a target relative to the lights 4.2.1. Horizontal measurement model. If the starting
in the horizontal and vertical directions when the platform position of target P relative to the retroreflector of the standard
moves step by step. This lowers the requirement for the high light beam OX is (x0, y0, z0 ), when the platform is moving, the
straightness of the horizontal granite guide rail and makes it standard light beams OX, OY and OZ have the respective travel
easy to design a long rail in large-scale length measurement. lengths LX, LY and LZ; and the retroreflector positions of OY
Using a pentaprism to make the light have a right angle turning and OZ are HY and HZ relative to the retroreflector position of
can give the plane two orthogonal axes, so it is not necessary OX; then the deflection angles α in the Y direction and β in the
to keep two rails exactly perpendicular to each other. Z direction are as follows:

 LX − LY

tan α =
4.2. Plane measurement models HY
(3)

 L X − LZ
Figure 4 shows the arrangement of retroreflectors and a tan β = .
HZ
pentaprism. A target p(x, y, z) is set on the carrier. According Set the centre of rotation of the platform on the
to the movements of the pneumatic working platform and retroreflector position of OX, when the platform deflects by
target carrier, the system can be used in three ways. Keeping a tiny angle α (it means the platform turns around angle α
the carrier fixed on the vertical rail and moving the platform, on the Z axis); according to the coordinate rotation rule, the
the system can be employed for horizontal measurement; position change of target P in the X (horizontal) direction will
keeping the platform immovable on the horizontal rail and be
driving the carrier along the vertical rail, the system can x = x0 cos α − y0 sin α, (4)
be used for vertical measurement. When both carrier and
platform have movement, the system will perform coordinate and the relative displacements x will be
measurement. x = x0 (1 − cos α) + y0 sin α. (5)

1770
 A large-scale laser plane calibration system

P31
offset

P21
step
P1 P42
Figure 5. Offset of moving platform.
2m
As the platform has deflections in the Y and Z directions, P10 6m
so the movement PX of target P in the direction X after
compensating the Abbe errors will be
PX = LX − y0 sin α − z0 sin β − x0 (1 − cos α)
− x0 (1 − cos β). (6)
As x0 is much smaller than the platform length (about
500 mm), neglecting the second-order tiny displacements Figure 6. The set-up position for calibrating a laser tracker.
x0 (1 − cos α) and x0 (1 − cos β), PX will be given as follows:
PX = LX − y0 sin α − z0 sin β. (7) 5. Calibration tests for a laser tracker

According the set-up position of a laser tracker, the dimensions

4.2.2. Vertical measurement model. As shown in figure 4, the of the laser plane are determined. When the target moves
position measurement of a target moving on the vertical axis on the rails, the tracker and plane system will record the
(Z direction) is realized by an interferometer (standard light
movements of the target. To transform the tracker records
beam OXZ) and a pentaprism N. For target P set approximately
into the plane coordinate system, the RMS value of the least-
on the vertical axis, and the straightness of the vertical rail
squares transformation errors will be used as the standard
very small, it is acceptable to neglect the influence of the Abbe
uncertainty of the tracker measuring a point. Changing the
errors. Keeping the horizontal platform stable, the movement
set-up position of the tracker, the RMS value will change
PZ of target P in the Z direction will be equal to the beam OXZ
too. Figure 6 shows the set-up position for calibrating a laser
travelling length LXZ:
tracker.
PZ = LXZ . (8) Let N1 , N2 , . . . , Nn+1 be the records of point P in the plane
system; P1 , P2 , . . . , Pn+1 the records in the tracker system; R
4.2.3. Coordinate measurement model. Unlike single axis the least-squares transformation matrix; and V1 , V2 , . . . , Vn+1
measurement in which the travel length of point P is only the transformation errors; here Vi = Ni − RPi . The standard
influenced by the straightness of the guide rail, the coordinate uncertainty mp of the tracker measuring a point will be
measurements of P in the plane are also affected by the 
n T
platform offsets relative to the standard beams (as shown in i=1 Vi Vi
figure 5). mp = ± . (11)
n−1
If every movement LXi of the platform is less than a step,
then the offsets Yoffset and Zoffset in two directions (Z and Y) can By changing the tracker set-up position more than twice,
be given by using deflection angles αi and βi . the fixed part a and changeable part b of mp can be given.
  Set d1 and d2 as different set-up distances, and c1 and c2 as
Yoffset = LXi tan αi
 (9) corresponding standard deviations (the RMS value of the least-
Zoffset = LXi tan βi . squares transformation errors); then
Set the pentaprism on the platform as N (x0 , y0 , 0) relative 
c1 = a + bd1
to the standard light OX, and the travel length of OXZ as LXZ; (12)
c2 = a + bd2 ,
then the movement length NX of pentaprism N in the direction
X will be which means
NX = LX − y0 sin α − x0 (1 − cos α) − x0 (1 − cos β) d 1 c2 − d 2 c1 c 1 − c2
a= and b = . (12)
and the displacement ZP of target P relative to the pentaprism d1 − d2 d1 − d2
N in the direction Z will be Generally, d1 and d2 should have different values: one
ZP = LXZ − NX . short for checking the fixed part a of the standard uncertainty
The movement PY of target P in the Y direction is caused mp and another more than double the length for checking the
by the deflection and offset of the platform: changeable part b.
Some tests are carried out by the developed plane system;
PY = Yoffset − x0 sin α − y0 (1 − cos α).
a tracker is set up as shown in figure 6. The first set-up distance
So the relative coordinates (PX , PY , PZ ) of target P will is about 2 m and second is about 6 m; the size of the plane used
be  is about 4 m × 1 m. Let the target move along the trail of ‘ ’
PX = NX − (ZP + z0 ) sin β in the plane; take the horizontal steps as about 450 mm and
PY = Yoffset − x0 sin α − y0 (1 − cos α) (10) vertical steps as about 100 mm; then more than 40 points are

PZ = ZP + Zoffset + x0 sin β. recorded. Repeat this three times in every situation; the RMS

1771
L Ma et al

Table 1. The RMS values of the least-squares transformation errors. structure, some factors which reduce the accuracy of a laser
Set-up Transformation plane can be eliminated; they are listed as follows.
distance (m) No RMS (mm) • Two light axes of the plane are made up by one beam of
1 0.020 light, so they are on the same plane.
2 2 0.021 • Using a pentaprism to create a right angle turning, makes
3 0.020 two light axes of the plane orthogonal.
1 0.031 • To measure the horizontal length with three
6 2 0.030 interferometric beams, compensates the Abbe errors
3 0.027 introduced by the rail and offsets of the target relative to
the light beams.
• Laser sources and most optic mirrors are set on a stable
platform at the end of the horizontal granite rail, so the
values of the least-squares transformation errors in every test
influences of environment on them are the same.
are shown in table 1.
The results of tests show that the plane system can evaluate Employing a laser plane meets the requirement of the standard
a tracker very well by using a RMS value of the least-squares B89.4.19 to evaluate a tracker’s performance, and can assess
transformation errors. the coordinate measurement uncertainty of the tracker. It is
Using equation (12) and taking the influence factor k = 2, an ideal way to carry out laboratory calibration for industrial
it is known that this tracker’s coordinate uncertainty is about large-scale spherical coordinate measuring instruments, and
U = (30 + 5L) µm (K = 2); L (unit: m) is the distance of a make it possible to trace a tracker to the wavelength.
target.
References
6. Conclusion
[1] Schwenke H, Härtig F, Wendt K and Wäldele F 2001 Future
The developed system can perform as a coordinate assessment challenges in co-ordinate metrology: addressing
based laboratory for a large-scale portable spatial coordinate metrological problems for very small and very large parts
Proc. IDW Workshop 2001 (Knoxville, TN, May) pp 7–10
measuring instrument, such as a laser tracker. Based on the [2] Swyt D A, Phillips S D and Palmateer J W 2001 Developments
concept of minimum plane, it is easier to establish a standard at NIST on traceability in dimensional measurements Proc.
plane for calibrating a tracker. By using a special optic SPIE 4401 245–52

1772