Laser ranging: a critical review of usual

techniques for distance measurement

Markus-Christian Amann Abstract. We review some usual laser range finding techniques for in-
Technische Universität München dustrial applications. After outlining the basic principles of triangulation
Walter Schottky Institut and time of flight [pulsed, phase-shift and frequency modulated continu-
D-85748 Garching, Am Coulombwall ous wave (FMCW)], we discuss their respective fundamental limitations.
Germany Selected examples of traditional and new applications are also briefly
presented. © 2001 Society of Photo-Optical Instrumentation Engineers.
Thierry Bosch [DOI: 10.1117/1.1330700]
Marc Lescure
ENSEEIHT-Laboratoire d’Electronique Subject terms: distance measurement; laser ranging; triangulation; three-
dimensional vision; laser radar.
2 Rue Charles Camichel
F 31071 Toulouse Paper DDM-10 received July 24, 2000; accepted for publication July 24, 2000.
France
E-mail: [email protected]

Risto Myllylä
University of Oulu
Department of Electrical Engineering and
Infotech Oulu
P.O. Box 4500
90014 Oulu
Finland

Marc Rioux
National Research Council, Canada
Institute for Information Technology
Visual Information Technology
Ottawa, Canada, K1A 0R6

1 Introduction In this paper, some laser range-finding techniques for

Absolute distance measuring devices are at an important industrial applications are briefly presented: triangulation,
pulsed time-of-flight, phase-shift measurement and
evolutionary stage in industrial countries in terms of appli-
frequency-modulated continuous wave 共FMCW兲 modula-
cations such as nondestructive testing, reverse engineering
tion. All these approaches are still further developed be-
predictive maintenance or virtual reality. They have the po- cause the choice of the technique to be used depends
tential to improve the productivity of firms or the quality of mainly on the required application. Moreover, each of them
manufactured products. In particular, noncontact distance presents limitations, which are discussed here.
measurements of rough targets are of great interests for
many industrial applications.
2 Optical Triangulation for 3-D Digitizing
The basic principle of active noncontact range-finding
devices is to project a signal 共radio, ultrasonic or optical兲 The desire to capture shape by optical means dates back to
onto an object and to process the reflected or scattered sig- the beginning of photography.2 In the 1860s François Vil-
nal to determine the distance. If a high resolution range- lème invented a process known as photosculpture, which
finder is needed, an optical source must be chosen because used 24 cameras. Profiles of the subject to be reproduced
radio and ultrasonic waves cannot be focused adequately. were taken on photographic plates, projected onto a screen
In addition to absolute distance measurement, laser range- 共using the magic lantern兲, and transferred to a piece of clay
using a pantograph.
finding devices are traditionally used for 3-D vision, di-
Commercial applications developed rapidly; studios
mensional control, positioning or level control. opened in Paris, London, and New York. They stayed in
Optical distance measurement methods can technically operation from 1863 to 1867, when it was realized that the
be put into three categories: interferometry, time-of-flight photosculpture process was not more economical than the
and triangulation methods.1 Considerable progress has been traditional way of doing sculpture. In any case, the process
achieved during recent years by understanding the basic still required a lot of human intervention. Indeed, the pho-
physical and information theoretical principles of range tosculpture process supplied only 24 profiles, which were
sensing. It appears that the advance in the design of lasers, used to rough down the piece of clay, leaving a large
integrated optics devices, emitter and receiver electronics amount of manual work to be done to finish the sculpture.
will lead to further interesting developments. A professional sculptor was needed and the photosculpture

10 Opt. Eng. 40(1) 10–19 (January 2001) 0091-3286/2001/$15.00 © 2001 Society of Photo-Optical Instrumentation Engineers
Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

process required quite an investment in terms of cameras,

projection and reproduction systems, and skilled labor to
operate them. It is only with the advent of computers that
the process has regained substantial interest, more than 100
years later.
The surface shape of objects can be imaged and digi-
tized using the following basic components: a light source
to define a specific pixel共s兲, such as an encoding-decoding
process 共e.g., triangulation, fringe and pattern projections,
and time of flight兲, a sensing device composed of a collect-
ing lens and a photodetector that converts light energy to
electrical signal, an analog-to-digital converter, and finally
a computer to process, display, and store the raw data.3
The following reviews the basic concepts behind optical
triangulation for 3-D digitizing applications. It emphasizes
the critical elements of optimal design and the limitations
related to the use of coherent light. It is shown, for ex-
Fig. 1 Image resolution of a diffraction-limited optical 3-D laser
ample, that speckle noise is a fundamental limit for the scanner for four typical laser wavelengths.
position sensing of laser spot centroid. This has an impact
on the choice of the position sensor geometry. Design
guidelines are also presented. 2.2 Optical Triangulation Geometry
On scattering, the spatial coherence of the laser light is lost,
2.1 Light Source which means that the depth of field used at the projection
can be useful only if the lens aperture is closed down at the
Conventional light sources can and are used for optical tri-
collection. Otherwise the focused laser spot is imaged as a
angulation, but laser sources have unique advantages for
blurred disk of light on the photodetector. A solution to this
3-D imaging. One of these is brightness, which cannot be
problem is to modify the conventional imaging geometry to
obtained by an incoherent emitter. Another is the spatial
conform to the Scheimpflug condition.
coherence that enables the laser beam to ‘‘stay in focus’’
Essentially this geometry enables the photodetector sur-
when projected on the scene. Nevertheless, this property is
face to ‘‘stay in focus’’ with the projected laser light. Its
limited by the law of diffraction, which is written here as
construction is very simple when the focal planes of the
the propagation 共along the z axis兲 of Gaussian laser beams.
lens are used. Indeed, it is known that a point on the pro-

冋 冉 冊册
jection axis located at f 共point 1兲 will be imaged at infinity
␭z 2 1/2
共Fig. 2兲. Consequently the inclination angle of the photode-
␻ 共 z 兲 ⫽ ␻ 0 1⫹ , 共1兲
␲␻ 20 tector is defined by drawing a line between that point and
the principal point of the collecting lens 共line 1兲. Similarly,
and, defining depth of field D f using the Rayleigh criterion on the other side of the lens, one knows that a point at
gives: infinity will be imaged at a distance f from the lens. The
inclination and position of the photodector are then ob-
tained by constructing a line 共line 2兲 parallel to line 1 pass-
D f ⫽2 ␲␻ 20 /␭, 共2兲
ing through the point 2.
which shows that the depth of field D f is larger when the 2.3 Fundamental Limits
laser wavelength ␭ is small 共toward the blue兲 and/or when
the laser beam spot size ␻ 0 is large. A detailed analysis is Sampling the shape along the X and the Y axes is usually
given in Ref. 4. done in a straightforward manner 关time or space interval,
Figure 1 shows a graph of the preceding equation when depending on the geometry of the photodetector共s兲兴. Sam-
a pixel 共voxel兲 is sampled every ␻ 0 in the X, Y, and Z
directions. On the horizontal axis are the physical dimen-
sions of the ‘‘volume of view’’ along the X, Y, and Z di-
rections, assuming that the volume of view is a cube. Ver-
tically are the number of pixels 共voxels兲 resolved along the
three-axis. A few typical laser wavelengths are shown. This
graph can be used as a guide when designing an optical
system for imaging shapes. It shows the maximum number
of pixels 共in terms of information transfer兲 that can be ex-
tracted from a specific volume. As an example, for a cube
that is 50 cm on a side, one has access to more than 2000
resolved spots in each direction axis X, Y, and Z using
visible wavelength laser light, which correspond to a cubic
element 共voxel兲 of 200 ␮m on the surface of the illumi-
nated object. Fig. 2 Geometric construction of the Scheimpflug condition.

Optical Engineering, Vol. 40 No. 1, January 2001 11

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

Fig. 4 Geometry for simultaneous recording of shapes and colors.

2.4 Color
A very simple way to add color to a digitized shape is
illustrated in Fig. 4, which is essentially the same optical
triangulation geometry as in Fig. 2, with two modifications.
Fig. 3 Speckle noise limits the position sensing resolution. Here the single wavelength laser is replaced by a RGB laser
or a multiwavelength laser consisting of arbitrary colors in
arbitrary numbers. As an example, an Ar-Kr gas laser emits
pling along the third dimension 共Z axis兲 often requires im- simultaneously more than 10 colors in the visible part of
age pattern centroid location and interpolation. This is the spectrum. Each can be used to make color measure-
where coherence shows its limitation. Indeed, because of ments. This is the modification at the projection level.
the coherent nature of the laser projection, the imaged laser The second modification is made at the collection level.
spot on the photodetector共s兲 is corrupted with speckle A prism 共or a wedge兲 is used to disperse the multiwave-
noise. Here again, the geometry of the optical system and length laser beam into multiple beams, each having an am-
the wavelength of the light are the parameters. plitude related to the ‘‘color’’ of the illuminated point on
Figure 3 shows how speckle noise adds to the uncer- the surface of the object. An interesting feature of this ge-
tainty of an imaged laser spot. The origin of the modulation ometry is that the shape and color are digitized simulta-
observed in the profile is related to the scattering of a pure neously, enabling a perfect registration of geometric and
wavelength when diffusion occurs at the surface of the ob- color data.
ject.
2.5 Conclusion
Within the projected laser spot, each scatterer can be
regarded as a coherent emitter, and because there are many Digitizing shapes requires careful design consideration. For
of them within the area of illumination the resulting image all active approaches including optical triangulation, the
is the coherent sum of spatially incoherent sources. The most critical element in order to achieve high-resolution
result is a random modulation multiplying the expected imaging is the projection system, not the collecting system.
smooth light profile. In Ref. 5, the relationship between the In fact, in some cases it is found that optimum perfor-
geometrical parameters is established as: mances are obtained where the photosensor elements are of
larger sizes. This is quite the opposite of the current trend
␴ 2 ⫽␭ 2 f 20 /2␲ 共 ␾ cos ␤ 兲 2 , 共3兲 in conventional 2-D imaging. Applications of 3-D digitiz-
ing cover a wide spectrum from industrial design, manufac-
where ␴ 2 is the position variance, ␭ the wavelength of the turing, simulation, inspection, and visual communication.
laser light, f 0 is the lens-CCD distance on the optical axis,
␾ is the aperture of the collecting lens, and ␤ is the Sche- 3 Laser Pulse Time-of-Flight Distance
impflug angle. Measurement
When typical values are given for each parameter, the 3.1 Introduction
centroid uncertainty is found to be of the order of a few
micrometers, and more interesting, the physical dimensions The laser pulse time-of-flight 共TOF兲 distance measuring
of the photosensor elements have no effect on that limit. technique was originally used in military and surveying ap-
Consequently, the larger the photosensors, the better are the plications. It refers to the time it takes for a pulse of energy
performances of the 3-D digitizer in terms of sensitivity, to travel from its transmitter to an observed object and then
speed, and depth of view. In practice, using wider photo- back to the receiver (t d ). If light is used as energy source,
sensor elements means that the interpolation is done over a the relevant parameter involved in range counting is the
wider dynamic range. As an example, if the speckle noise speed of light 共roughly c⫽30 cm/ns兲. A TOF system mea-
limits is at 2 ␮m and the photosensor width is at 16 ␮m, sures the round trip time between a light pulse emission and
one requires only 5 bits of numerical interpolation. On the the return of the pulse echo resulting from its reflectance
other hand, if the photosensor width is 50 ␮m, 7 bits of off an object. Using elementary physics, distance is deter-
interpolation are required. Notice that at least 2 bits are mined by multiplying the velocity of light by the time light
added to the ratio to reduce quantization 共or numerical兲 takes to travel the distance. In this case, the measured time
noise to a negligible level. is representative of traveling twice the distance and must,

12 Optical Engineering, Vol. 40 No. 1, January 2001

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

Fig. 5 Block diagram of TOF laser range finder.

Fig. 6 Schematic of focal plane scanning.

therefore, be reduced by half to give the actual range to the

target. To obtain 1 mm accuracy, the accuracy of the time In 3-D measurements, the laser range finder is equipped
interval measurement should be 6.7 ps. with angle encoders to enable the definition of the coordi-
Since a single pulse is adequate for the unequivocal de- nates of the measurement point. Scanning is mechanical
termination of distance with centimeter precision and accu- and is carried out manually or automatically. In some ap-
racy depends only weakly on distance, this method is par- plications, manual scanning is adequate but for time-critical
ticularly appropriate, for example, in applications involving purposes a servo system is required to increase the mea-
distances longer than 1 m, applications where reflectors are surement rate. Basic techniques include scanning either the
not used and fast measurement applications such as scan- measuring head or only the measuring beam by means of
ning. In addition, averaging enables millimeter or even sub- galvanometer-driven mirrors.
millimeter precision to be achieved. The advantage of the Focal plane scanning can be used instead of a narrow
TOF system arises from the direct nature of its sensing as laser beam, which scans a surface mechanically, point by
both the transmitted and returned signals follow essentially point. This enables a range map to be obtained without
the same direct path to an object and back to the receiver. mechanical beam scanning. The result is highly improved
Some new applications such as sensors in robotics, au- 3-D mapping performance, particularly with respect to
tonomous vehicles and rendezvous and docking, along with measuring time, at considerably reduced mechanical com-
anticollision and proximity sensors and sensors used in pro- plexity and thus reduced size and power requirements. The
tecting an area in front of a machine demand limitations on principle of focal plane scanning is presented in Fig. 6.
instrument size, mass and power consumption. To achieve The laser beam illuminates the total field of view on the
these goals, the basic building blocks of a TOF range finder surface. A target’s surface is viewed using matrices of
have to be realized in the form of high-performance inte- separate detectors. Each detector covers its own fraction of
grated circuits. Bosch and Lescure provide an excellent re- the field of view illuminated by the laser. Detector signals
view of absolute distance measurement in one of the SPIE are analyzed in the time domain, and distances to particular
Milestone volumes.1 points are calculated on the basis of time interval measure-
ments. The system can simultaneously measure distances to
3.2 Constructions several directions without any moving parts.6
A pulsed TOF distance measuring device consists of a laser 3.3 Timing Jitter and Walk, Nonlinearity and Drift
transmitter emitting pulses with a duration of 5 to 50 ns, a
receiver channel including a p-i-n or an avalanche photodi- The main sources of inaccuracy in laser rangefinders are
ode, amplifiers, an automatic gain control 共AGC兲 and tim- noise-generated timing jitter, walk, nonlinearity and drift.
ing discriminators. The emitted light pulse 共start pulse兲 trig- Typical noise sources include noise generated by the elec-
gers the time interval measurement unit, and the reflected tronics, shot noise caused by the background radiation-
light pulse 共stop pulse兲 stops it. The distance to the target is induced current and shot noise created by the noise of the
proportional to the time interval. A block diagram of a laser signal current. Jitter in timing determines mainly the preci-
range finder is shown in Fig. 5. sion of the range measurement. The amount of timing jitter
The selection of laser type depends on the intended mea- is proportional to noise amplitude 关root mean square 共rms兲兴
surement range and the required speed. For long distances and inversely proportional to the slope of the timing pulse
共up to several kilometers兲, a Nd:YAG laser can be used, at the moment of timing (du/dt). A single-shot resolution
giving peak powers extending to the megawatt level. Low- of 1 cm can typically be achieved with a good signal
priced pulsed laser diodes 关single heterostructure 共SH兲 or (SNR⫽100) using the 100 MHz bandwidth of the receiver
double heterostructure 共DH兲-type兴, capable of producing channel. However, precision deteriorates as the distance in-
peak powers of tens of watts, enable measuring distances creases and the pulse amplitude decreases proportional to
up to a few hundreds of meters—or even longer using co- the square of the distance.7 Pulse amplitude and shape
herent summing. The repetition frequency of YAG lasers is variations create timing error in the time-pickoff circuit and
low, whereas laser diodes can be used at rates of tens of that error is called walk error. Jitter and walk in leading
kilohertz, the DH-type may even reach the megahertz level. edge timing are shown in Fig. 7.

Optical Engineering, Vol. 40 No. 1, January 2001 13

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

The final precision of the distance measurement can be

greatly improved by averaging, with the improvement be-
ing proportional to the square root of the number of results
averaged. Thus, by averaging 100 successive measure-
ments, the final resolution can be improved to the millime-
ter level, the corresponding measurement time being 1 ms
with a pulsing rate of 100 kHz, for instance.
If the statistical error is averaged to a negligible level,
Fig. 7 Timing jitter and walk.
the accuracy of the system is defined by its systematic er-
rors such as nonlinearity in the time interval measurement
scale and drift. A careful design of the system can reduce
The time discriminator is a very important part of a pre- these errors to the millimeter level.
cision time measurement system. The task of the discrimi-
nator is to observe time information from the electric pulse
of the detector preamplifier and to produce a triggering sig-
nal at the right instant. The choice of time derivation 3.4 New Applications and Development Trends
method depends on the desired time resolution, counting Several new applications for the transit time measurement
rate and required dynamic range of the pulse. Commonly of laser pulses are currently being developed to supplement
used principles in discriminator design include leading traditional ones. A promising method involves using the
edge timing 共constant amplitude兲, zero crossing timing path length of light pulses in human tissue, pulp and paper
共derivation兲, first moment timing 共integration兲, and constant or optical fiber as a sensor principle. In this context, the
fraction timing. Constant fraction discrimination 共CFD兲 term photon migration is often used to describe the propa-
compensates with idealized pulse shapes for walk caused
gation of light in scattering media like human tissue. In
by both amplitude and risetime and is commonly used in
turbid media, photons take a number of different paths
the TOF measuring units of laser range finders.
The principle behind the operation of CFD is the search thereby broadening short light pulses. The use of TOF tech-
for an instant in the pulse when its height bears a constant niques for imaging soft tissues 共optical tomography兲 is be-
ratio to pulse amplitude. The occurrence of this point pro- ing actively investigated by many researchers today for ap-
duces a triggering pulse. The constant fraction instant can plications such as breast cancer diagnosis and imaging the
be examined with a high gain differential emitter-coupled oxygenation state of the brain in newborn infants.11 From
logic 共ECL兲 comparator, which amplifies the difference of the point of view of the paper industry, one of the most
the attenuated and the delayed pulses.8 CFD compensates important properties of graphic paper is the interaction of
for walk caused by amplitude and rise time, but not for light and paper structure. One method of obtaining a more
walk caused by unlinear shape variations. Zero crossing detailed understanding of the propagation of light in paper
and first moment timing compensate for amplitude varia- and pulp is the high-resolution measurement of the delay
tions, while leading edge timing fails to compensate for any light pulses suffer as they pass through a sheet of paper or
variation described. a sample of pulp.12 As for composite materials, optical fi-
The transmitter, gain-control and time-interval measure- bers can be embedded in them during the manufacturing
ment units are also critical for the accuracy of the system. process. Calculations are then based on the fact that the
The transmitter should be able to produce a stable laser TOF of the light pulse in a fiber is a function of the length
pulse shape. This may call for the temperature stabilization and refractive index of the fiber, which are affected by
of the diode. The dynamic phenomena of the laser diodes, stress, temperature and pressure.13
relaxation oscillations, should also be considered when de- Commercial applications in the civilian sector place sev-
signing the pulsing scheme as they may easily lead to sig- eral demands on laser-based devices. First, the so-called
nificant changes in laser pulse shape. class 1 laser condition 共eye-safety兲 that limits the peak
Gain control is needed to adjust the dynamics of the
power of the laser to a few watts should be fulfilled. How-
timing pulse to that of the timing discriminator. The
ever, the reliable detection of low-reflectance or high-
amount of control needed depends on the measurement
temperature targets requires a sufficiently high optical peak
range, construction of the optics and reflectivity variations
of the object. An adjustable optical attenuator at the re- power. The use of picosecond pulses helps to overcome the
ceiver optics can be used to realize gain control. The ad- eye-safety problem, but at the cost of a receiver with high
vantage of this method over electrical gain control is its bandwidth. Shorter pulses also give better precision.14
delay stability over a wide control range. Recently, the dy- Other highly desirable qualities of the range finder include
namic range of electronic gain control methods has in- cost-effectiveness. The size, weight and power consump-
creased quite remarkably.9 tion of the device should be reduced to increase its potential
The time interval between the start and stop pulses is application range. These goals can be achieved using full
measured with the time-to-digital converter 共TDC兲, which custom application-specific integrated circuits 共ASIC兲. A
is a fast, accurate and stable time-interval measuring device long-term vision could be to realize the TOF range finder as
that uses, e.g., a digital counting technique together with an a component-like microsystem where all the basic elements
analog or digital interpolation method.10 The single shot 共laser diode, photodetector, receiver channel and time inter-
resolution of the TDC is typically better than the noise val measurement electronics兲 are located on a single encap-
generated timing jitter. sulated hybrid circuit.

14 Optical Engineering, Vol. 40 No. 1, January 2001

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

where

P 0 ⫽ power output of the laser diode

T T ⫽ collection and transmission efficiency of
transmitter
T R ⫽ transmission of the receiver optics
␳ d ⫽ Lambertian reflection coefficient of the target
␪ ⫽ laser beam incident angle on the target
A R ⫽ area of receiver lenses.
Fig. 8 Block diagram of a phase-shift laser rangefinder using a het-
erodyne technique.
Only, the Lambertian component is used for the calcu-
lation of the photoelectric signal; the specular component is
ignored in Eq. 共5兲. As a matter of fact, the term ␳ d P 0 / ␲ is
4 Principle and Limitations of a Laser the power intensity 共in watts per steradion兲 reflected per-
Phase-Shift Range Finder pendicular to the target surface.
For some representative construction materials 共such as
4.1 Introduction standard printing white paper, unpainted wood, unpolished
The principle of a laser phase-shift range finder is described coating, satin-like coating兲 the values of ␳ d are between 0.6
in this section. Heterodyne technique gives a narrow band- and 0.7 for a wavelength ␭⫽820 nm 共Ref. 15兲. With ␭
width amplification which improves the SNR. The limita- ⫽950 nm, values can vary from 0.4 to 0.8 共Ref. 16兲.
tions such as high level of the photoelectric signal, inter-
mediate frequency drift and electrical crosstalk are 4.4 Error Analysis
discussed.
4.4.1 High levels on the rf channel
4.2 Physical Principle of a Phase-Shift Rangefinder The power budget shows that the ratio P R / P 0 may vary
with a factor greater than 1000 if the distance D varies for
In a phase-shift rangefinder, the optical power is modulated example from 1 to 10 m. Thus, when the photoelectric sig-
with a constant frequency. The basic operating scheme of nal amplitude varies with a factor 1000, the phase-shift
the device is shown in Fig. 8. A sine wave of frequency f rf error ␦␸ introduced by this variation must be less than 0.1
generated by the main oscillator modulates the dc current deg for an error distance measurement ␦ D of 2.5 mm. But,
of the laser diode. After reflection from the target, a photo- with large amplitude of the useful signal, distortion and
diode collects a part of the laser beam. Measurement of the clipping introduce phase-shift errors. To avoid this error,
distance D is deduced from the phase shift ⌬ ␸ ⫽2 ␲ f rf⌬t the first solution is the defocusing of the photodiode, so that
between the photoelectric current and the modulated emit- it receives less light when measured distances are small. A
ted signal: second solution is increasing the feedback of the mixer to
avoid distortion and clipping.17
1 ⌬␸
D⫽ c f , 共4兲
2 2 ␲ rf 4.4.2 Intermediate frequency drift
With a low-power laser diode, a narrow bandwidth ampli-
where c is the speed of light in free space and ⌬t is the fication improves the SNR. For a white noise at the input,
TOF. the output of a passband second-order filter presents a
When ⌬ ␸ ⫽2 ␲ , the unambiguous distance measure- noise-equivalent bandwidth B n given by the relationship18:
ment is limited to ⌳⫽c f rf /2 共for example, ⌳⫽9 m with
f rf⫽16.66 MHz兲. To ameliorate the accuracy of this setup, B n ⫽ 共 ␲ /2兲 ⌬ f rf . 共6兲
the phase shift is not directly measured at the working high
frequency but at an intermediate frequency f rf⫽ 兩 f rf⫺ f ol兩 Thus, the rms noise of the useful signal is proportional
using a heterodyne technique that preserves the phase shift to the square root of the bandwidth ⌬ f if . Because of sym-
versus distance. The signals of the two mixers outputs are metry, it can be supposed that the frequency drift ␦ f rf or
filtered by a passband circuit tuned on f if and with a ⌬ f if ␦ f ol of one oscillator balances the drift of the other one. No
bandwidth. error is present if the two intermediate frequency filters are
identical on reference channel and signal channel. But if the
value of the bandwidth is too narrow, we must take into
4.3 Received Signal Power account the device mismatch effects introduced in interme-
When the target area is as large as the area illuminated by diate frequency tuned circuits by the frequency drifts of f rf
the transmitter, the received signal power P R is given by and f ol . For example, for Rauch filter structure, the quality
the relationship: factor Q⫽ f if /⌬ f if and the tuned frequency f if mismatches
are imposed by the presence of capacitance mismatches.
␳d AR With f rf⫽16.66 MHz, ⌬ f if⫽200 Hz, assuming a capaci-
P R ⫽T T T R P cos ␪ 2 , 共5兲
␲ 0 D tance mismatch of 1%, to achieve a phase-shift error

Optical Engineering, Vol. 40 No. 1, January 2001 15

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

smaller than ␦ ␸ ⫽0.1 deg 共distance error ␦ D⫽2.5 mm兲, the receiver channel. A large p-i-n photodiode surface gives
required stability of the two main oscillators is ␦ f rf / f rf such a field of view. So, the large photodiode capacitance
⬇3.5⫻10⫺7 . increases the time constant at the input of signal channel,
However, to guarantee the resolution, the best method that reduces the bandwidth and/or the phase margin of the
consists of keeping the value of the intermediate frequency transimpedance amplifier. One solution is to compensate
constant by a phase-locked loop technique. In Refs. 19 and the photodiode capacitance by a tuned circuit. This tech-
20, 100 m distance measurement to a retroreflector is ob- nique is possible because the phase-shift range finder is
tained 共i.e., with a good SNR兲 with errors of less than 50 working at one frequency f rf only.
␮m. The stable f rf⫽1.5 GHz modulating frequency is gen-
4.5 Conclusion
erated from the 15th harmonic of a 100 MHz output of a
Rubidium atomic clock. Distance measurement by the phase-shift technique is a
good method to obtain a resolution of some millimeters in 1
4.4.3 Influence of the crosstalk to 20 m ranges with noncooperative targets. For coopera-
tive targets 共cube corner兲, the resolution can be better than
Electrical crosstalk between the transmitter and the receiver 50 ␮m, with high-frequency modulation. Because the pho-
produces another important error. When the photoelectric toelectric current is a sine wave, a photodiode with a large
signal and the modulation current of the laser diode have area can be used to achieve a wide angle of the field of
the same frequency f rf , the synchronous leakage v crosst , view. In this way, it is possible to scan the laser beam by
arising from the modulation current source, is superim- micromirrors, enabling compactness and low cost of future
posed on the working photoelectric signal v ph . The two 3-D vision system.
signals are added vectorially. The transimpedance amplifier
associated with the photodiode must be shielded to exclude
5 FMCW Optical Radar
undesired external signals. However, because the influence
of the crosstalk cannot be totally removed with shielding 5.1 Introduction
techniques, different methods have been proposed. To re-
move the effects of this field leakage, the gain switch of a Optical distance measurement with the FMCW technique
laser diode is used as a light comb generator, and an ava- has been used in various applications such as noncontact
lanche photodiode 共APD兲 selects the second harmonic of surface profiling, fiber optic sensing, reflectometry, posi-
the photoelectric signal.21 Another method, based on the tioning and tomography. The interest in the FMCW tech-
use of a Pockels cell, has been proposed to obtain a mea- nique is due to the large dynamic range and the high reso-
sured signal at a frequency that is a multiple of the modu- lution, particularly at short range sensing. Because of the
lation frequency, although its implementation is still rather recent progress in the area of laser diode technology, high-
performance FMCW ranging systems can currently be re-
difficult.22
alized with electronically tunable laser diodes.25–29 In this
The crosstalk errors at the modulation frequency f rf de- paper, we therefore describe the operation principle of the
termine the minimum signal-to-induction ratio necessary to FMCW technique and discuss the system performance
obtain a given accuracy. For example, to limit the maxi- achievable with laser diodes.
mum error phase shift ␦␸ to 0.1 deg, the amplitude ratio of
the photoelectric signal on the leakage signal must be 600
共55 dB兲. Because the induced signal is ‘‘masked’’ by the 5.2 Principle
electrical noise, the SNR must be higher than 55 dB. This The principal setup of an optical FMCW radar system is
condition is difficult to satisfy when the frequency modu- illustrated in Fig. 9. The optical power from a frequency
lation f rf of a low-power laser diode is higher than 10 MHz. modulated laser diode, the instantaneous frequency of
The maximum distance measurement error ␦ D max is inde- which is periodically shifted by ⌬ f , is used as probing
pendent from the modulation frequency and increases with signal as shown in Fig. 10共a兲.
the square of the distance, in accordance with Eq. 共5兲. Be- The periodic and linear frequency chirp may practically
cause the photoelectric current and crosstalk are added vec- be performed by applying a saw-tooth bias current to the
torially, a periodic error is obtained versus distance D. In modulator section of the wavelength-tunable laser diode.
this way, this error can be compensated.23 The laser output passes an optical isolator to avoid deterio-
To determine a rough estimation of the crosstalk be- riations of the laser frequency by reflections and is then
tween the driver laser and the photodiode, a model of fic- sent simultaneously to the object and the reference mirror,
titious sources of perturbation at the input of the transim- and the reflected signals are then superimposed in a square
pedance amplifier has been proposed.24 The coupling is law detector diode. Owing to the detection process that is
represented by the mutual capacitance and the mutual in- proportional to power, i.e., the amplitude squared, both sig-
ductance. These models show the advantage of using an nals are mixed in the detector and the main ac component
APD. Indeed, the gain of the primary photoelectric current of the electrical output is at the frequency difference f if of
occurs inside the semiconductor crystal itself, and it can be the two optical signals. The detector output is fed into an
assumed that the crosstalk becomes superimposed on the amplifier-limiter so that unintentional amplitude modula-
output signal only when the signal has been amplified. But, tion is suppressed. Finally, the intermediate frequency f if of
as APD areas are small, this gives a small field of view of the reflected signals is measured with a frequency counter.
the receiver. For 3-D vision, if only the laser beam is de- Due to the square law mixing process the amplitude of the
flected by micromirrors, a large field of view of the receiver detector output at f if is proportional to the amplitudes 共not
is necessary, when no scanning mirrors are used in the the powers兲 of the object signal and the reference, respec-

16 Optical Engineering, Vol. 40 No. 1, January 2001

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

Fig. 9 Principle setup of an FMCW optical radar system using a

tunable laser diode as light source.

tively. Accordingly, the dynamic range of the FMCW tech- Fig. 11 Schematic setup of an improved FMCW optical radar sys-
nique is twice as large than that of the pulse radar tech- tem with a reference line to eliminate the effect of a nonlinear optical
nique, where the electrical signal is proportional to the frequency ramp.
object signal power.
The range difference between object and reference mir-
ror R, which is the relevant quantity to be determined, is ␶ ⫽2R/c, 共7兲
proportional to f if as indicated in Fig. 10共a兲. Since the
round-trip delay time ␶ of the object signal is given as where c is the light velocity, the intermediate frequency
amounts to

f if⫽⌬ f ␶ /t m ⫽2⌬ f R/ct m , 共8兲

where t m denotes the ramp period, which is typically of the

order 0.1 to 1 ms. Accordingly, the distance sensing is done
by an electric frequency measurement 共usually in the kilo-
hertz regime兲, while in the case of the more popular pulse
radar, the delay time is measured directly. Since the ramp
period can be chosen arbitrarily, the FMCW radar can de-
termine ␶ values in the picosecond range, according to mil-
limeter distances R, by simply performing a frequency
measurement in the kilohertz regime. Consequently, no
high-speed electronic is required to determine delay times
even in the subpicosecond range, according to micrometer
distances R occurring, e.g., in the automated inspection of
printed boards.

5.3 Limitations
Unfortunately, the frequency modulation response of a la-
ser diode is, in general, nonuniform against the modulation
frequency, so that a linear optical frequency sweep cannot
be realized by a linear modulation of the control current. In
addition, the frequency versus control current characteristic
is also in general nonlinear. As a consequence, deviations
Fig. 10 Instantaneous optical frequencies versus time of the object from the linear ramp as shown in Fig. 10共b兲 usually occur
and reference mirror optical signals (a) in case of a linear optical
frequency ramp, the intermediate frequency f if is constant between
that, in turn, lead to a variation of the intermediate fre-
t 1 and t 2 and (b) in case of a nonlinear optical frequency ramp, a quency f if as shown schematically in Fig. 11. While the
chirping intermediate frequency occurs. total phase difference ␾ as indicated by the shaded areas in

Optical Engineering, Vol. 40 No. 1, January 2001 17

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

Figs. 10共a兲 and 10共b兲 is still proportional to ␶ and the dis- This brief review of laser range-finding techniques holds
tance R as long as ␶ Ⰶt m , the variation of f if during each out the promise of an exciting future for traditional and new
ramp increases the total intermediate frequency bandwidth industrial applications.
and, consequently, the noise level that finally limits the
accuracy. A usual technique to eliminate the effect of the References
nonlinear frequency ramp is to use simultaneously a refer- 1. T. Bosch and M. Lescure, Eds., Selected Papers on Laser Distance
Measurement, SPIE Milestone Series, Vol. MS 115, SPIE Optical
ence line, as shown in Fig. 11, so that the ratio of distance Engineering Press, Bellingham, WA 共1995兲.
R and the length of the reference line L ref is independent of 2. B. Newhall, ‘‘Photosculpture,’’ Image 7共5兲, 100–105 共1958兲.
the linearity of the frequency ramp.30 Various schemes 3. M. Rioux, F. Blais, J.-A. Beraldin, and P. Boulanger, ‘‘Range imaging
sensors development at NRC laboratories,’’ in Proc. IEEE Workshop
have been presented to linearize the optical frequency on Interpretation of 3D Scenes, pp. 154–160, Austin, TX 共1989兲.
sweep of the laser diodes, that commonly utilize a reference 4. M. Rioux, G. Bechthold, D. Taylor, and M. Duggan, ‘‘Design of a
large depth of view three-dimensional camera for robot vision,’’ Opt.
delay line similar to Fig. 11 to determine the variations of Eng. 26共12兲, 1245–1250 共1987兲.
the intermediate frequency f if,1 . By means of an electrical 5. R. Baribeau and M. Rioux, ‘‘Influence of speckle on laser range find-
feedback path the control current ramp to the laser is ad- ers,’’ Appl. Opt. 30共20兲, 2873–2878 共1991兲.
6. R. Myllylä, J. Marszalec, J. Kostamovaara, A. Mäntyniemi, and G.-J.
justed such that f if,1 is kept constant.31 As a consequence, Ulbrich, ‘‘Imaging distance measurements using TOF lidar,’’ J. Opt.
the current ramp deviates from linearity in such a way that 29, 188–193 共1998兲.
7. K. Määttä, J. Kostamovaara, and R. Myllylä, ‘‘Profiling of hot sur-
the relevant instantaneous optical frequency versus time faces by pulsed time-of-flight laser range finder techniques,’’ Appl.
characteristic becomes linearized according to Fig. 10共a兲. Opt. 32共27兲, 5334–5347 共1993兲.
The maximal measurable distance of an FMCW laser 8. M. R. Maier and P. Sperr, ‘‘On the construction of a fast constant
fraction trigger with integrated circuits and application to various pho-
diode system is finally limited by the coherence length of totubes,’’ Nucl. Instrum. Methods 87, 13–18 共1970兲.
the laser diode. The finite value of the latter, which typi- 9. T. Ruotsalainen, P. Palojärvi, and J. Kostamovaara, ‘‘A current-mode
gain-control scheme with constant bandwidth and propagation delay
cally is of the order of several tens of meters is a conse- for transimpedance preamplifier,’’ IEEE J. Solid-State Circuits 34共2兲,
quence of the laser diode phase noise that also determines 253–258 共1999兲.
the spectral linewidth. Correspondingly, laser diodes with 10. E. Räisänen-Ruotsalainen, T. Rahkonen, and J. Kostamovaara, ‘‘Inte-
grated time-to-digital converters based on interpolation,’’ J. Analog
narrow linewidths within the frequency sweep range are Integr. Circ. Signal Process. 15共1兲, 49–57 共1998兲.
best suited for the measurement of longer distances. Re- 11. A. J. Joblin, ‘‘Method of calculating the image resolution of a near-
cently, however, it was shown that this limitation can be infrared time-of-flight tissue-imaging system,’’ Appl. Opt. 35共4兲,
752–757 共1996兲.
overcome by application of a frequency-shifted feedback 12. J. Carlsson, P. Hellentin, A. Malmqvist, W. Persson, and C.-G. Wahl-
laser and range measurements up to 18 km were reported.32 ström, ‘‘Time-resolved studies of light propagation in paper,’’ Appl.
Opt. 34共9兲, 1528–1535 共1995兲.
Also the ultimate limitation of the measurement accu- 13. H. Lahtinen, M. Jurvakainen, P. Pramila, H. Tabell, M. Kusevic, O.
racy is due to the phase noise of the laser diode.33,34 Again, Hormi, V. Lyöri, M. Heikkinen, R. Myllylä, P. Suopajärvi, and H.
frequency-modulated laser diodes with narrow spectral Kopola, ‘‘Utilisation of optical fibre measurement techniques in de-
termination of residual stresses of composites,’’ Proc. SPIE 3746,
linewidth are in general preferable for highly accurate 526–529 共1999兲.
range measurements. However, it should be noted that the 14. G. Kompa, ‘‘Powerful picosecond pulsed laser radar with micrometer
low-frequency phase noise occurring in the f if frequency ranging resolution,’’ in Proc. 26th Eur. Microwave Conf., pp. 147–
152 共1996兲.
range is relevant. Strictly speaking, therefore, in case of 15. T. Bosch and M. Lescure, ‘‘Experimental determination of the useful
nonwhite phase noise the laser linewidth represents not a reflection coefficient of non-cooperative targets for a time-of-flight
laser rangefinder,’’ Opt. Rev. 2共B04兲, 289–291 共1995兲.
useful quantity to estimate the relevant phase noise. This 16. D. Nitzan, A. Brain, and R. Duda, ‘‘The measurement and use of
particularly holds for the widely continuously tunable laser registered reflectance and range data in scene analysis,’’ Proc. IEEE
65共2兲, 206–220 共1977兲.
diodes that exhibit large ⌬ f values of the order of several 17. G. Perchet, M. Lescure, and Th. Bosch, ‘‘Errors analysis of phase-
hundred gigahertz with the correspondingly large sensitiv- shift laser range finder with high level signal,’’ Sens. Actuators A
ity. Experimentally, high resolution in the submicrometer 62共1–3兲, 534–538 共1997兲.
18. E. Cherry and D. Hooper, Amplifying Devices and Low-Pass Ampli-
regime has been demonstrated with diode laser FMCW fier Design, Wiley, New York 共1968兲.
ranging systems.28 Using widely continuously tunable laser 19. J. M. Payne, D. Parker, and R. F. Bradley, ‘‘Rangefinfer with fast
multiple range capability,’’ Rev. Sci. Instrum. 63共6兲, 3311–3316
diodes with ⌬ f of 500 GHz, a spectral linewidth of 25 共1992兲.
MHz, and a repetition period t m of 10 ␮s, the one-shot 20. M. A. Goldman, R. E. Creager, D. H. Parker, and J. M. Payne,
relative accuracies for distances of 1 m and 1 cm, respec- ‘‘Rangefinder metrology for the Green bank telescope,’’ in Proc. EOS
Topical Meeting on Optoelectronic Distance/Displacement Measure-
tively, amount35 to 4.3⫻10⫺5 and 2.7⫻10⫺4 . These results ments and Applications, 共1997兲.
clearly prove the high performance of this technique for the 21. K. Seta and T. Oh’ishi, ‘‘Distance measurement using a pulse train
emitted from a laser diode,’’ Jpn. J. Appl. Phys., Part 2 26, L1690–
accurate and fast measurement of short distances and, con- L1692 共1987兲.
sidering the large modulation bandwidth of laser diodes, 22. M. Lescure, T. Bosch, and A. Dziadowiec, ‘‘An electro-optical fre-
make this technique also well suited for 3-D viewing sys- quency doubling method to remove electrical noise in laser range
finder,’’ IEEE Trans. Instrum. Meas. 40, 1046–1047 共1991兲.
tems. 23. R. Müller, H. Wölfelschneider, M. Kublin, and H. P. Anderer, ‘‘Ap-
plication limits of optical distance measuring based on the phase de-
tection technique,’’ in Proc. EOS Topical Meeting on Optoelectronic
6 Conclusion Distance/Displacement Measurements and Applications, 共1997兲.
24. Th. Bosch and M. Lescure, ‘‘Error analysis of 1–10 m laser phase-
Usual laser distance measuring techniques were briefly re- shift rangefinder,’’ IEEE Trans. Instrum. Meas. 46共6兲, 1224–1228
ported. This paper presented the basic principles of trian- 共1997兲.
25. G. Beheim and K. Fritsch, ‘‘Remote displacement measurements us-
gulation and TOF 共pulsed, phase-shift, and FMCW兲 and ing a laser diode,’’ Electron. Lett. 21, 93–94 共1983兲.
some related applications. It also emphasized their funda- 26. E. M. Strzelecki, D. A. Cohen, and L. Coldren, ‘‘Investigation of
tunable single frequency diode lasers for sensor applications,’’ J.
mental limitations and proposed solutions to improve their Lightwave Technol. LT-6, 1610–1680 共1988兲.
performances in terms of accuracy, sensitivity and cost. 27. A. R. Slotwinski, F. E. Goodwin, and D. L. Simonson, ‘‘Utilizing

18 Optical Engineering, Vol. 40 No. 1, January 2001

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms
 Amann et al.: Laser ranging: a critical review of usual techniques . . .

AlGaAs laser diodes as a source for frequency modulated continuous Thierry Bosch: Biography appears with the guest editorial in this
wave 共FMCW兲 coherent laser radars,’’ Proc. SPIE 1043, 245–251 issue.
共1989兲.
28. E. C. Burrows and K.-Y. Liou, ‘‘High-resolution laser LIDAR utiliz-
ing two-section distributed feedback semiconductor laser as a coher- Marc Lescure received the PhD degree in
ent source,’’ Electron. Lett. 26, 577–579 共1990兲. electronic engineering from the University
29. A. Dieckmann, ‘‘FMCW-LIDAR with tunable twin-guide laser di- of Toulouse in 1972, and the Doctorat
ode,’’ Electron. Lett. 30, 308–309 共1994兲.
30. A. Dieckmann and M.-C. Amann, ‘‘Phase-noise-limited accuracy of d’Etat in 1985. Since 1972, he has been
distance measurements in a frequency-modulated continuous-wave working at the Electronic Laboratory
LIDAR with a tunable twin-guide laser diode,’’ Opt. Eng. 34, 896– (ENSEEIHT) of the National Polytechnics
903 共1995兲. Institut of Toulouse. From 1972 to 1985,
31. K. Iiyama, L-T. Wang, and K. Hayashi, ‘‘Linearizing optical his research included characterization of
frequency-sweep of a laser diode for FMCW reflectometry,’’ J. Light- optoelectronic devices (LED, solar cell,
wave Technol. 14, 173–178 共1996兲. MIS) by frequential method. Since 1986,
32. K. Nakamura, T. Hara, M. Yoshida, T. Miyahara, and H. Ito, ‘‘Optical his teaching/research interests has in-
frequency domain ranging by a frequency-shifted feedback laser,’’
IEEE J. Quantum Electron. 36, 305–316 共2000兲. cluded design of analog circuits, wireless infrared communications
33. G. Economou, R. C. Youngquist, and D. E. N. Davies, ‘‘Limitations and optoelectronic sensors. He has published two books with A.
and noise in interferometric systems using frequency ramped single- Dziadowiec on the subject of analog circuits and he is coeditor with
mode diode lasers,’’ J. Lightwave Technol. 4, 1601–1608 共1986兲. Thierry Bosch of Selected Papers on Laser Distance Measurements
34. M.-C. Amann, ‘‘Phase-noise limited resolution of coherent LIDAR in the SPIE Milestone Series.
using widely tunable laser diodes,’’ Electron. Lett. 28, 1694–1696
共1992兲.
35. A. Dieckmann and M.-C. Amann, ‘‘FMCW-LIDAR with tunable Risto Myllylä received his DSc (engineer-
twin-guide laser diode,’’ in Trends in Optical Fibre Metrology and ing) degree in electrical engineering from
Standards, O. D. D. Soares, Ed., pp. 791–802, Kluwer Academic the University of Oulu, Finland, in 1976. He
Publishers, Dordrecht 共1995兲.
has been an associate professor with the
Department of Electrical Engineering, Uni-
Markus-Christian Amann received his versity of Oulu since 1974 and a professor
Diplom degree in electrical engineering in since 1995. From 1977 to 1978, he was a
1976 and his DrIng degree in 1981, both visiting scientist with the University of Stut-
from the Technical University of Munich. tgart, Germany. From 1988 to 1995 he was
During his thesis work, he studied superlu- a research professor with the Technical
minescent diodes and low-threshold laser Research Centre of Finland. His research
diodes and developed the AlGaAs-GaAs interests include industrial and biomedical instrumentation develop-
metal-clad ridge-waveguide laser. From ment, particularly in optical measurements. He has coauthored one
1981 to 1994 he was with the Corporate book and over 240 papers and patents. Dr. Myllylä is a past presi-
Research Laboratories of the Siemens AG dent of Finnish Optical Society, a board member of European Opti-
in Munich where he was involved in the re- cal Society, and a member of SPIE, OSA, IEEE, and EOS.
search on long-wavelength InGaAsP-InP laser diodes. In 1987 he
became the head of a research group working on advanced laser
diode structures and integrated optoelectronic devices. In February Marc Rioux is a principal research officer
1994 he joined the Department of Electrical Engineering at the Uni- with the Institute for Information Technol-
versity of Kassel, establishing a working group for III/V semiconduc- ogy at the National Research Council. His
tor electronics and optoelectronics. Since November 1997 he has present interests are in the development of
held the chair of semiconductor technology at the Walter Schottky 3-D digitizing, modeling and display for ma-
Institute of the Technical University of Munich, where he is currently chine vision, optical dimensional inspection
engaged in research on tunable laser diodes for the near IR, quan- and visual communication in the Visual In-
tum cascade lasers, long-wavelength vertical-cavity laser diodes formation Technology Group. He received
and laser diode applications. Amann has authored or coauthored his BS degree in engineering physics in
about 120 papers on semiconductor optoelectronics in scientific 1971 and his MS in physics in 1976, both
journals and conference proceedings and books. Together with from Laval University. He worked for 5
Jens Buus he recently wrote a book on tunable laser diodes. He is a years on CO2 laser development and applications and 2 years on IR
member of the German Informationstechnische Gesellschaft (ITG), holography and joined the National Research Council in 1978 to
and a senior member of the IEEE Lasers and Electro-Optics Soci- work on optical sensor research.
ety.

Optical Engineering, Vol. 40 No. 1, January 2001 19

Downloaded From: http://opticalengineering.spiedigitallibrary.org/ on 09/14/2012 Terms of Use: http://spiedl.org/terms