Sie sind auf Seite 1von 10

Roll angle in 6DOF tracking

Stephen Kyle
Senior Honorary Research Fellow, University College London
Honorary Research Fellow, West Midlands Manufacturing Measurement Centre, UK

Abstract
In a 3D polar measurement system which
tracks a probing device in 6 degrees of
freedom (6DOF), the probes roll angle can be
identified as a critical parameter in some
system design concepts. This paper reviews
methods and proposals for roll angle
measurement in single laser tracker systems
based on existing implementations and
published concepts.
Introduction
An earlier CMSC paper (Kyle, 2006)
highlights roll angle as a critical measurement
in a 6DOF single laser (polar) tracking system.
Roll angle appears, directly and indirectly, in a
number of patents, for example recently by
Leica Geosystems (2007), which emphasizes
its importance in commercial manufacture.
There are currently 3 manufacturers of laser
tracking systems, i.e. Hexagon (Leica
Geosystems), API and Faro, and the first two
now offer real-time probing. Since the tracked
position is offset from the measured point of
interest, e.g. the ruby ball on a touch probe or
the laser spot on a non-contact probe or
scanner, the probe must be tracked in 6DOF in
order to calculate the offset vector from
tracked to measured point.
Standard single laser tracking measures the 3D
location of a target retro-reflector by range and
two angles (approximately horizontal and
vertical).
Roll

The 3 angular elements of the target, here


identified as roll, pitch and yaw angles (see
Figure 1), can be measured as a combined
result from single image processing. Two
methods are well represented here:
1. Leicas T-System uses a zoom camera
image of targets on the probing device
which surround the reflector. The
technique
is
essentially
a
photogrammetrists space resection in
which the range data is discarded, see
Loser and Kyle, 2003.
2. Past research at the Technical University
of Vienna (no longer active) projected the
return beam onto a CCD chip. The angular
orientation could be determined from the
shadow image of the reflector edges, see
Prenninger et al., 1993.
Alternatively, two of the angles, pitch and
yaw, can be separately measured to a relatively
high accuracy. Consider, for example, a probe
based on a hand-held camera sighting a single
target on the tracker. An up/down rotation or a
side-to-side rotation will correspondingly
move the target image up and down or side to
side. Its image position in x, say, is therefore a
measure of yaw angle and in y is a measure of
pitch angle. Given a normal angle of view, for
example 50 which is roughly equivalent to the
acceptance angle of a reflector, a 2K x 2K
imaging chip and 1/20 pixel interpolation on
the image, then accuracy for a well calibrated
image could approach 5 arc sec, equivalent to
25m at 1m offset. This is a good accuracy for
a large volume metrology (LVM) probe.
Unfortunately, roll is still missing in this case.

Pitch

Yaw

This document will therefore review methods


and proposals for roll angle determination as a
largely separate task from pitch and yaw
measurement.

Figure 1 Aircraft roll, pitch and yaw angles


CMSC: Charlotte-Concord, July 21st. 25th. 2008

Page 1

Stephen.Kyle@micronSpace.com

Introduction: the space triangle

Figure 2 Theodolite orientation (6DOF)

A standard method for orienting non-levelled


theodolites, i.e. determining their relative 6
degrees of freedom (6DOF), helps in
understanding
the
geometry
of
roll
measurement, see Figure 2. Each theodolite
points at the other (details below), thus
obtaining vectors R1 and R2. Relative to
theodolite 1, theodolite 2 is then fixed in 3D by
R1 and the separation D (see below). Vector
R2 provides the pitch and yaw of theodolite 2.
By rolling theodolite 2 such that its measured
vector V2 to an offset target intersects the
equivalent offset target vector V1 from
theodolite 1, roll is also fixed. Instruments and
target form a space triangle.
For accurate scale the theodolite separation D
must be measured in some way. Theodolites
typically derive scale, and hence D, by
measuring a second offset target on a scale bar
of length s. However, D, or equivalently d1 or
d2, can also be measured separately or directly
by substituting a Total Station or laser tracker
(see below), for either theodolite.

instruments represent a stationary tracker and a


moving target probe, it will clearly be a
convenience to confine all measurements to a
region close to the baseline between them, i.e.
an offset target at locations (1) or (3). As a
further convenience, the target would be
attached to its nearby device and be part of it.
In this case the attached instrument cannot
directly measure the corresponding offset
vector, shown in red, and it must be
determined by prior calibration. Further
discussion will identify the shaded space
triangle as a roll plane and the offset target as a
roll target. Note also that each device measures
two vectors within the roll plane.
The location of the offset target highlights an
accuracy issue. For the same pointing
accuracy, roll is less accurately determined as
the offset target moves closer to the baseline
and when it is on the baseline the roll angle
cannot be determined at all (the roll plane
disappears in this case). There are two general
configurations for accurate roll determination:
Lo-res/Hi-offset
A device with moderate angle resolution
requires a large offset angle A, close to 90.
In a triangulation scheme involving
theodolites, cameras or a mixture of both,
aiming for the optimal offset would be a
common strategy.
Hi-res/lo-offset
A device with high angle resolution can
accommodate a small offset angle A. This
is an appropriate strategy, if possible, for
6D single laser tracking.

Figure 3 Generalized procedure

The procedure is generalized in Figure 3,


which shows two angle measuring devices
(yellow circle and triangle) and 3 possible
positions of an offset target. If the measuring
CMSC: Charlotte-Concord, July 21st. 25th. 2008

Figure 4 Reciprocal pointing by theodolite


Page 2

Stephen.Kyle@micronSpace.com

There is an interesting aspect to the theodolite


orientation method. To obtain the pointings
between theodolites it is common to sight
offset targets on the theodolite telescopes, as
indicated in Figure 4. The vector mean of R1a
and R1b, for example, generates the required
pointing R1. (Some older Wild theodolites had
internal targets visible through the telescope
lens with an apparent location at the
theodolites centre of rotation. Direct sighting
along the baseline was therefore possible, but
this was not a common solution.)
As can be seen, this process generates several
roll planes which could then be used for
calculating the relative roll angle, assuming the
local offset vectors (in red) have been
calibrated as indicated earlier. As far as the
author is aware this is never actually done with
theodolites. This is the lo-res/hi-angle
requirement and a more accurately defined roll
plane is obtained by sighting a more offset
target, typically on a scale bar, this
measurement being required in any case.
However, once a polar measuring device, such
as a laser tracker, is substituted for a
theodolite, the situation changes.
The relevance of the discussion above is
illustrated by Figure 5 which extends an early
concept suggested by Kyle, 1991. The original
suggestion was to have a motorized theodolite
track a moving camera.

theodolite. Alternatively, scale is found by


locating a second offset target on a scale bar of
length s.
By adding a stylus and probe ball to the
camera it becomes a touch probe for 3D object
measurement, provided the probe ball location
within the cameras coordinate system is
determined by prior calibration.
Some of this early work included methods of
targeting the cameras perspective centre and
theodolites rotation centre so that the vectors
between them could be directly identified. A
very simple way of doing this is possible with
modern tracker technology, as follows.
2
2.1

Roll plane by offset targeting


Roll target in tracker space

Figure 6 Pinhole retroreflector

Figure 7 Multiimaging pinhole


reflector

Figure 5 Camera tracked by (video) theodolite

The main space triangle is shown. Scale is


found by a range measurement, d, to a fixed
offset target, possibly by using an add-on
EDM or Total Station in place of the
CMSC: Charlotte-Concord, July 21st. 25th. 2008

A type of camera whose perspective centre can


be directly sighted or measured is available to
laser trackers and shown in Figure 6. This
concept appears in patents by both Boeing,
2002 and Leica, 2003. Typically the apex of a
prism retro-reflector is removed to create a
small hole. Part of the incoming laser beam,
most of which is reflected back on itself,
passes through the hole and onto a sensor. If
this is a CCD, or similar, then the xy chip
coordinates of the laser spot are a measure of
the direction of the laser beam (or equivalently
the pitch and yaw of the prism). This is, in
effect, a pinhole camera and retro-reflector
combined in one.
Page 3

Stephen.Kyle@micronSpace.com

Leicas later 2007 patent utilizes the device for


roll measurement, an example being shown in
Figure 7. The tracker follows the pinhole
camera/reflector which effectively measures
the pointing back to the tracker. An offset,
luminous target on the tracker is also imaged
by the pinhole camera/reflector, creating the
roll plane indicated by the shaded space
triangle. Comparison with Figure 5 shows that
the essential difference is direct range
measurement in the 2007 patent compared
with indirect range measurement in the
tracking theodolite concept.
This concept is a lo-res/hi-angle configuration,
i.e. the angle subtended at the pinhole camera
should be relatively large in order to optimize
roll angle accuracy. For the convenient
situation where targets are placed on or near
the tracker head, this will be the case when the
probe is close to the tracker but not when it
moves further away. Multiple roll targets
distributed away from the tracker offer one
way to improve the situation at longer ranges
but more attention must then be given to
ensuring lines of sight are maintained. In this
case, in order to avoid the need for many
targets such that one is always visible to the
reflector as it is moved around by the operator,
robot etc., a motorized pointing could be
employed to maintain an optimal view. In fact,
the Boeing patent is designed to do exactly this
for the more specific condition of maintaining
a good pointing back to the tracker only. APIs
recently developed Active Target is a
commercial device which also does this.
Clearly an actively pointed pinhole prism
could be controlled to maintain multiple
pointings and more optimal roll.

The combination of pinhole camera + retroreflector can also be replaced by a


conventional camera with an attached, offset,
standard retro-reflector. This is used in a
6DOF tracking concept patented by Metronor,
2006 and shown in Figure 8. A retro-reflecting
tracker target, attached to a camera with
probing stylus, is located in 3D. The camera
images a minimum of two targets, shown
green and known in the trackers coordinate
system, for example by prior direct
measurement by the tracker. Alternatively, by
choosing a target arrangement which directly
identifies the trackers origin of measurement,
for example as illustrated by the 4 blue targets
which define lines intersecting at the required
position, then only one offset target would be
required and the geometry would be closer to
the Leica configuration.
The full 3D offset of the reflector from the
perspective centre in the cameras coordinate
system, shown as a red line, must be known.
However, this configuration has excess
information to make the 6DOF calculation.
The trackers range measurement, D, is not
mathematically required although it will
considerably enhance accuracy in many
geometrical configurations 1 .

Figure 9 Alternative camera tracking

Figure 9 shows the concept with tracking range


measurement removed, and it should be
1

Figure 8 Metronor concept


CMSC: Charlotte-Concord, July 21st. 25th. 2008

In fact, Leicas 2007 concept, if using a 3D


calibration of offset target position, also has excess
information. However, it would not be practical to
take scale from such a short offset. In fact, the
offset need not be calibrated in full 3D and only
direction calibration data is required.
Page 4

Stephen.Kyle@micronSpace.com

compared with Figure 5. The main difference


from the earlier concept is that the camera
sights two offset targets rather than one offset
target and the theodolite. As with the laser
tracker concept, the targets must be known
with respect to the tracking instrument. One
option amongst others is to use a robot Total
Station which can only track real-time in angle
but can locate the targets with range
measurements d1 and d2 in an initialization
procedure.
Now that direct measurement between device
centres has been removed, the geometry is a
little more complex than that of a simple space
triangle, but not difficult. Using a geomatic
engineers standard horizontal resection it is
easy to calculate the circle through the location
of the camera and two targets by using the
known positions of the targets and the
subtended angle measured at the camera. In 2D
the camera then lies somewhere on this circle.
When the circle is rotated about the chord
connecting the targets, a doughnut shape is
created which is the surface of all possible 3D
positions of the camera, each with a particular
angular orientation. The pointing from the
theodolite intersects this surface, thus fixing
the camera in 6DOF. It is a convenience for
the theodolite to use an offset target on the
camera, i.e. the blue target and pointing rather
than the direct pointing. If the local offset
vector is known in the cameras coordinate
system, then every point on the doughnut
surface has a corresponding offset point which,
in turn, creates a nearby surface for the
possible target positions. Again the
intersection of the theodolite pointing with this
offset surface locates the camera probe in
6DOF. In practice the calculations would be
done differently but the geometry illuminates
the concept. With minor modification, the
same geometrical description can also be
applied to Figure 5.

2.2

Roll target on the probe

Figure 10 Roll target on probe

Figure 11 Alternative probe roll

As explained earlier, the roll target can be at


either end of the system. Figure 10 shows a
concept presented by Kyle, 2005 in which the
overview camera of the API tracker could
potentially measure a full roll angle at the
reflector using a single offset target. The
shaded space triangle is only an
approximation here since the cameras
perspective centre is separated from the
trackers measurement origin. This 3D offset
must be known by prior calibration. Compare
this with Figure 12 (below) and note that the
single offset target is the minimum condition.
However, a more accurate roll angle would be
expected with at least one more offset target,
as implied by the additional greyed elements in
the diagram. Applying a substitution used
earlier, the pinhole retro-reflector can be
replaced by a camera/reflector combination, as
indicated in Figure 11. The camera on the
probe will require its own target on the tracker
for which again the 3D offset in local tracker
space is required. (It is interesting to compare
this with either pointing shown in Figure 4.)

This is again a lo-res/hi-angle configuration


which benefits from roll targets with greater
offsets and, potentially, a motor driven probe
which maintains optimal pointings.

Figure 12 Leica patent 2007


CMSC: Charlotte-Concord, July 21st. 25th. 2008

Page 5

Stephen.Kyle@micronSpace.com

With a relatively wide-angle, overview camera


this is the lo-res/hi-angle condition for optimal
roll and this will only be achieved at close
ranges. However, if an on-board zoom camera
is used, such as the T-Cam in the Leica TSystem (a vario zoom lens camera), then the
situation becomes hi-res/lo-angle and therefore
capable of operation at longer ranges for
equivalently good roll angles. This is shown in
Figure 12 where the patent specifies a
minimum of 2 targets and suggests alternative
pattern arrangements.

Figure 13 Roll by barcode

Some optimization might be possible by using


the type of barcode target common in digital
levelling, a suggestion in Kyle, 2006. A
circular barcode was suggested by the concept
diagram, repeated in Figure 13, although two
or more linear sections of barcode would also
work. The potential advantage is that pattern
matching might locate the barcode in the
image to a very small sub-pixel level, as is the
case in digital levelling, so enabling an
accurate roll angle to be determined.
3

Roll plane by tilt sensing

In non-levelled theodolite orientation, shown


in Figure 2, an offset roll target at infinity will
generate offset roll vectors which are parallel.
Using tilt sensors for roll determination can
therefore be viewed as a variant of the space
triangle method, discussed earlier.
Roll by tilt sensing has advantages
No additional target.
No offset roll target is required and so there
is no requirement for lines of sight off the
baseline.
Constant sensing accuracy to maximum
device separation, typically 40m for laser
trackers.
If device separation is largely horizontal,
the large offset angle improves roll
accuracy.
However, there are disadvantages:
There is a failure case when the baseline
between
the
devices
is
vertical.
The roll plane then disappears. When the
line is close to vertical the geometry is also
poor.
Restricted range
Tilt sensors offering an angle accuracy
suitable for metrology applications
generally have a limited operational range,
instead of the ideal 360 of roll.
Dynamic effects.
Acceleration distorts gravity sensing and
movement causes oscillations, both a
source of inaccuracy.
Potential ways to deal with restricted range and
dynamic effects are suggested in Kyle, 2005.

Figure 14 Levelled theodolite orientation

When levelled theodolites are oriented to one


another their local z axes are set parallel. In
fact, they can also be tilted and the local
direction of the gravity vector found by tilt
sensing or other means, see Kyle, 1990.

CMSC: Charlotte-Concord, July 21st. 25th. 2008

Figure 15 Tilt sensor attached to pinhole


reflector

Page 6

Stephen.Kyle@micronSpace.com

Figure 16 6DOF tracking from API

Verbal discussions with API indicate that tilt


sensing is used in their Intelliprobe and
SmartTrack sensors, and use of a single tilt
sensor would explain their limitation on the
roll angle. It is also not clear if there is tilt
sensing only at the target. If so, this would
seem to require a setup or calibration routine to
determine a datum position for the roll angle.
4

Roll plane by transmitted reference


direction
A different approach to measuring roll angle is
to manufacture a roll plane in some way.
Optical methods of transmitting a reference
direction between devices are presented here.
4.1 Plane of polarization.
A beam of polarized light, whose direction of
polarization can be detected by various means,
appears in many suggestions for transmitting a
roll orientation angle. The fact that polarized
laser beams are used in laser trackers makes
this technique an obvious candidate to
consider.

The idea of using polarimetry to detect roll


angle is a long-established concept and there
are numerous investigations. As an example,
King and Raine, 1981, suggest roll angle could
be measured by polarimetry over several
metres to a precision of 0.2 arc sec, or a few
arc sec over hundreds of metres. There are also
a number of patents in this area, such as the
one from Daimler Benz, 1985, and from Lau
(API), 2003. The latter specifically aims to
extract the roll angle from a tracking laser
beam. However, obtaining an accurate result in
real time, and with light and compact optics
and electronics, may be a challenge in practice
and a possible reason for APIs selecting
instead the simpler technology of tilt sensing.
4.2 Projected patterns: Bird system
This early 1983 concept is introduced as
background to the remaining discussion. Here
the motivation was to track the location and
pose of a robot within a large volume and to
this end a prototype device was presented by
Arai, et al., 1983.

Figure 18 Bird system concept

This is essentially a real-time triangulation


system in which two angle trackers project
cross-shaped laser beams onto CCD ring
sensors placed onto the faces of a moving
target cube.

Figure 17 Roll angle by polarized light

Figure 17 is taken from Kyle, 1990, and


illustrates the concept. Here it was suggested
that polarized light, transmitted from one
theodolite and analyzed by the other, could
determine the roll angle between their
reciprocal pointings.
CMSC: Charlotte-Concord, July 21st. 25th. 2008

It is apparent that the roll of the cross is


detected by where the arms intersect the sensor
ring, but pitch and yaw are also detected to
some extent when the cross is incident at an
oblique angle to the surface of the cube. For
example, if the arms are generated with a 90
intersection angle, they will, in general, be
detected on the cube as having some other
angle. (Note that pitch and yaw accuracy
would not be high. For example, in
Page 7

Stephen.Kyle@micronSpace.com

perpendicular viewing they cannot be detected


if they represent rotations about the arms of the
cross.) Two projected crosses provide
sufficient data to determine the cubes angular
orientation in space. The centre of the cross
determines a particular location on the cube,
thereby placing the cube on a particular line in
space with respect to one of the angle trackers.
The corresponding line from the second
tracker locks it in 6DOF. It is not necessary for
the lines to intersect. Some features of this
early system:
A tracking mechanism
It is provisionally assumed that since the
cube is located in 6DOF, the current
location in object space of the centres of the
ring sensors can always be computed and
returned to the trackers as a point which
they must follow.
3 CCD sensors were recommended in order
to measure a wide angle of rotation
The paper does not suggest multiple
trackers so it is again provisionally assumed
that the trackers would automatically
switch to tracking a different ring centre
when a change in the cubes orientation
blocked the line of sight to the currently
tracked ring.
Small sensors
The prototype ring sensors were small, only
about 4mm in diameter
From the viewpoint of modern laser tracking
technology this concept could be considerably
improved to provide a 3D/6D measurement
system.
A single polar device eliminates the need
for the second tracker
Accurate pitch and yaw measurement can
be made by pinhole reflector or reflector +
camera
A single line is then sufficient for roll
measurement but a pattern may also be
useful
Modern, large area and linear array sensors
would improve sensitivity
With this in mind, some more recent proposals
can be evaluated.

CMSC: Charlotte-Concord, July 21st. 25th. 2008

4.3

Projected lines and patterns

Roll plane
generator

Roll reference plane

Roll plane
detection

Figure 19 Added roll reference plane

Figure 19 illustrates a concept taken from


Leicas 2007 patent. An additional laser roll
reference plane is generated by a separately
mounted device on the tracker. Linear CCD
sensors on a roll ring around a pinhole prism
reflector detect the beam and hence define the
roll angle.
The roll reference plane can be constructed in
a number of ways, for example by expanding a
laser beam through a cylindrical lens or
reflecting it off rotating mirrors. When added
as a fixed ancillary device, the plane must
approximately cover the vertical operating
range of the tracker, indicated by the heavy red
dashed lines. However, it could project over a
more limited angular range, by being mounted
in a tilting unit like the T-Cam and, like the TCam, directed to maintain a pointing at the
target.
In this particular example, the reference plane
will be detected by one of the linear sensors in
any roll orientation, and only one measurement
is required. However, it is potentially more
accurate to measure the line of the plane at two
or more positions.

Page 8

Stephen.Kyle@micronSpace.com

Figure 20 Closed linear arrays for roll plane


detection

In fact, Arais ring sensor could be simulated


in a much larger and alternative form using
modern linear CCD arrays. This could then
replace the separated arrangement of arrays
above.
Assuming relative array positions can be
determined, the example arrangements in
Figure 20 show ways in which the roll plane
(green) could be detected in two locations to a
potentially much higher accuracy than in the
Bird system
There could be advantages in using crossed
planes to ensure an optimal intersection with
the linear arrays and/or an improved roll
accuracy through excess measurement.

Figure 21 Area array for roll detection

Figure 22 Back projected roll beam

Figure 23 Coaxial roll beam, semi-reflecting


target
CMSC: Charlotte-Concord, July 21st. 25th. 2008

Line or pattern projection onto area array


CCDs would clearly also work. Although
defining a much smaller sensing area than the
closed arrangements of linear arrays, they also
enable measurement of the entire pattern,
which might compensate for the lower
resolution. Of the many issues to be resolved
in a working system, two are highlighted.
1. Placement of sensor
These general approaches are developed
from Kyle, 2006. Figure 22 shows a
moving pan and tilt target device
projecting a roll pattern onto an offset area
array on the fixed tracker. The target
device tracks in a similar way to the Bird
system and the geometry is similar to
Figure 4. Alternatively, in Figure 23 a roll
beam (green), co-axial with the tracker
beam (red) is projected from the fixed
tracker onto the target device. Here beam
splitting takes the tracker beam through a
mask whose shadow projection measures
pitch and yaw and the orientation of the
pattern on the CCD measures roll.
2. Generation of the pattern
Although there are many systems currently
commercially available which can project
laser patterns, detailed investigation is still
required to see if they have a sufficient
range (focusable or otherwise) and are
sufficiently compact and cost effective to
be applied.
5 Conclusions
Separately and accurately measured roll angle,
when combined with an accurate measurement
of pitch and yaw, would enable high accuracy
6DOF laser tracking of object probing devices.
For the same spatial accuracy, improved
rotational accuracy permits longer offsets from
the tracked location to the probed point on the
object. Users will always want more from a
system, and a longer reach outside the line-ofsight could be regarded as a worthwhile
improvement.
For hand-held probes, further investigations
might focus on:
Projected patterns from tracker to probe
Optimized imaging at the tracker of
targeting around the reflector
All-angle tilt sensing with algorithm
compensation
of
accelerations
and
oscillations
Page 9

Stephen.Kyle@micronSpace.com

Motorized probes, likely to be heavier than


hand-held probes and therefore potentially
better applied to robot and machine control or
as components in CMM arms, might make use
of:
Polarized light
Back-projected roll planes (from probing
device to tracker)
The space triangle with large offset roll
targets

Kyle, S., 2005


Alternatives in 6DOF probing - more
flexibility, lower cost, universal?
Presented at CMSC, Austin, Texas, July 2005

6 Acknowledgements
The author is grateful to Leica Geosystems for
recognizing his contributions to concepts for
6DOF tracking whilst in their employment,
some of which have been incorporated into
their 2007 patent referenced below.

Lau, K., (Automated Precision Inc.), 2003


Six dimensional laser tracking system and
method
US Patent 20030043362

7 References
Aarai, Endoh, Minokoshi, 1983
Position and orientation measurement of a
moving object by CCD photo array sensors
Proceedings of 4th. Intl. Conference on
Assembly Automation, Tokyo, 1983, pp 133
144.
Boeing: Greenwood, T., 2002
Steerable retroreflective system
method.
United States Patent 6,420,694

and

Daimler Benz, 1985


Patent number DE3347833
Device
for
continuous
polarimetric
measurement of the angle of roll of a movable
machine part
King, R., J., Raine, K. W., 1981
Polarimetry applied to alignment and angle
measurement
Optical Engineering, 20 (1), 39 - 43, 1981.
Kyle, S., 1990
Orientation with polarimetry and gravity
vectors
Land & Minerals Surveying, Vol. 8, no. 3,
March 1990, pp 122 - 129.

Kyle, S., 2006


Optically jointed probing systems for large
volume coordinate metrology
The Journal of the CMSC, vol. 2, no. 1, Spring
2007.

Leica Geosystems: Loser, R., Markendorf,


A., 2003
Method and device for determining spatial
positions and orientations.
United States Patent 6,667,798 (also
WO0109642)
Leica Geosystems: Zumbrunn, R. et al.,
2007
Measurement system for determining 6
degrees of freedom of an object,
US patent: US 7,312,862 B2
Loser, R., and Kyle, S., 2003
Concepts and components of a novel 6DOF
tracking system for 3D metrology
Optical 3D Measurement Techniques VI, (Eds.
Grn, Kahmen), vol. 2, pp 55 - 62, 2003.
Metronor, 2006
Measurement of spatial coordinates,
US patent: US 7,145,647 B2
Prenninger, Vincze and Gander, 1993
Measuring dynamic robot movements in
6DOF and real time
Robot Calibration (Eds. Bernhardt, Albright),
Chapman and Hall 1993, Ch 8, pp 124 - 153.

Kyle, S., 1991


Reciprocal observations in photogrammetry
Photogrammetric Record, 13(77): 729 - 739
(April 1991).

CMSC: Charlotte-Concord, July 21st. 25th. 2008

Page 10

Das könnte Ihnen auch gefallen