Professional Documents
Culture Documents
www.elsevier.com/locate/robot
Received 20 July 2005; received in revised form 27 July 2006; accepted 4 August 2006
Available online 18 September 2006
Abstract
Rodents demonstrate an outstanding capability of tactile perception with their whiskers. Mechanoreceptors surrounding the whisker shaft
in their follicle structure measure deflection of the whisker. We designed biomimetic whiskers following the basic design of the follicle. In
experiments with the artificial whiskers, we have explored tactile perception based on active whisking where the deflection angle or velocity
provides the localization information which is the basis of shape recognition. Measuring contact distances at varying protraction angles allows
discrimination of round objects with a varying curvature, or objects with different lateral shapes, such as square and round objects. We show the
capabilities and limitations of a single whisker for shape recognition as well as the usefulness of multiple whiskers. In addition, measuring both
vertical and horizontal deflection of a single whisker allows detection of the vertical shape for objects with a smooth surface. Two or more whiskers
stacked vertically can recognize the vertical shape by observing the difference of their deflection amplitudes or the time shift of deflection velocity
peak. The results provide a clue on how autonomous robots could improve their sensory capabilities with mechanical probes.
c 2006 Elsevier B.V. All rights reserved.
Keywords: Biomimetic whisker; Biorobotics; Tactile sensors; Shape recognition; Active perception
2. Method
Fig. 2. Experiments with artificial whiskers (a) artificial whiskers on a Koala robot (four magnetic sensors and two microphone-based sensors at one side, and four
microphone-based sensors at front) (b) different types of objects formed by discs.
Fig. 3. Our active whisking system (a) active whisking and bending angle (the longitudinal axis of some whisker may not go through the centre of rotation) (b)
varying deflection distance with a given torque (if we take the reference axis from the center of rotation to the contact position, the angle β is the angle between
the surface slope and the reference axis, and the angle θd represents the angle of the whisker beam at contact position relative to the reference axis. The relation
between the two angles influences the deflection angle θ0 − θ1 ).
where tan θ is the slope at a position x (0 ≤ x ≤ d) along the a fixed contact location while the motor is rotating the support
whisker bar, starting at the clamped position, θ is measured with plate shown in Fig. 3(a).
an axis connecting from the whisker base to the contact point, The position of whisker base for a rotational angle θ0 is
and d is the distance of an object from the whisker base at the denoted by a vector cE with the centre of rotation as a reference
onset of contact. The magnetic sensors in the artificial whisker point — see Fig. 3. Then cE = (|E c| cos θ0 , |E
c| sin θ0 ) where
measure the deflection of the whisker 14 mm away from the c| is the distance between the centre of rotation and the
|E
whisker base. whisker base, and θ0 is measured relative to the axis from the
In our artificial whisker system displayed in Fig. 3, the center of rotation to the contact position. The whisker shaft is
whisker base (clamped position) is 45 mm away from the clamped at the whisker base and so the whisker base will be
rotational axis of the DC motor, and the position of the whisker alternatively called the clamped position, which provides one
base moves by the rotation of the DC motor. Now we define of the boundary conditions of the Bernoulli–Euler equation. We
the term contact distance as the distance to the contact position also define pE for a position vector on the contact point (| pE| will
from the center of rotation, and deflection distance as the be the distance from the centre of rotation to the contact point,
distance to the contact point from the whisker base. Thus, the simply denoted as p in the text), and the motor axis will be the
contact distance will be constant for a fixed contact position, but reference point. Then | pE| and d = | pE − cE| represent contact
the deflection distance changes in a sweeping period even with distance and deflection distance, respectively. If we take the
232 D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243
Fig. 4. Theoretical deflection amplitudes depending on contact distance (with the assumption that the whisker bar is long enough to be bent with the corresponding
distance) (a) varying distances (γ = 0) (b) varying γ ’s (160 mm distance).
reference axis from the centre of rotation to the contact position, where the angles θ0 , θ1 and α change during the protraction
the angles at the whisker base and at the sensor position relative period, but γ is constant. From the above analysis for our
to the axis will be (θ0 + α), (θ1 + α), respectively, where whisker design, the deflection angle is inversely proportional to
α = cos−1 [ pE · ( pE − cE)/(| pE| · | pE − cE|)]. The instantaneous the contact distance as shown in Fig. 4(a), and smaller distances
torque at the whisker base depends on the distance between the increase the deflection angle more rapidly in proportion to
clamped position and the contact location. Thus, Eq. (1) can be protraction angle.
re-formulated as below: To determine the slope at contact position in active whisking
τ 2 1 (θd in Fig. 3), we can substitute x = 0 and x = d in Eq. (2).
E I tan(θ + α) = x − τ x + τ d. (2) Then we obtain
2d 3
By applying the sensor position x = h = 14 mm and the 1
tan[θd (t) − γ + α(t)] = − tan[θ0 (t) − γ + α(t)] (5)
clamped position x = 0 to the above equation, we can cancel 2
out τ and obtain the following equation from E I tan[θ0 (t) − γ + α(t)] = τ d/3 and E I tan[θd (t) −
2
3h 3h
γ + α(t)] = −τ d/6, where θd (t), θ0 (t) is the angle at contact
tan(θ1 + α) = − + 1 · tan(θ0 + α) (3) position and clamped position at time t, respectively (θ0 is set
2d 2 d
to 0 at the onset time of contact).
where h is the distance 14 mm between sensor position and
clamped position, and λ = θ0 − θ1 is the deflection angle or 2.2. Shape recognition
bending angle that the sensors measure, which is distinguished
from the protraction angle θ0 . The angle α between the two We now apply our biomimetic whiskers to the problem of
vectors, cE and pE − cE, is updated for each protraction angle recognizing the shape of a target object. To estimate slope or
θ0 . The above Eq. (3) implies that the bending angle alone curvature of an object, the whiskers need to touch many contact
is sufficient to estimate the contact distance1 as well as the points around the object. Theoretically, we can estimate the
deflection distance. shape of an object by collecting a set of distance values while
The longitudinal axis of a whisker may not go through the
the whisker is sliding over the surface of an object. This may
center of rotation. That is, the angle between cE and pE − cE at the
need free angular control of the whisker beam in 3-dimensional
onset time of contact may not be zero. Then the angle can be
space.2 In our experiments, we will test a restricted domain
defined as |γ | = cos−1 [Ec · ( pE − cE)/(|E
c| · | pE − cE|)]. The angle
of shape recognition based on just sweeping movements in
γ is the rotational angle of cE counterclockwise relative to the
the horizontal plane. We will observe the capabilities and
reference vector pE − cE, and we obtain θ0 + α = γ at the onset
limitations of a single whisker for shape recognition and also
of contact. With γ 6= 0, θ0 and θ1 should be replaced by θ0 − γ
evaluate the usefulness of multiple whiskers. First, we will
and θ1 − γ in Eq. (3), respectively. Then we have
investigate discriminating a round object and an edged-surface
2
object, which will be called lateral shape recognition. Second,
3h 3h
tan(θ1 − γ + α) = − + 1 · tan(θ0 − γ + α) (4)
2d 2 d
2 Rodents use horizontal-plane sweeping to recognize the shape or texture
1 The contact distance | pE| can be calculated as | pE| = d cos(α) + |E
c| cos(θ0 ) with a series of head movement sequences in which increasingly refined
where |E
c| is 45 mm in our case. information is collected [5,11].
D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243 233
Fig. 5. Round and square objects and their change of contact position (α1 is the angle protracted before contact, and it determines the angular position of an object).
Fig. 7. Deflection signal for a cube (a) deflection amplitude (b) deflection velocity.
Fig. 8. Deflection signals depending on contact distance (top: deflection amplitude, bottom: deflection velocity) (a) square object for point contact (b) round object
with diameter 100 mm (c) square object with side length 50 mm and a side angle β = 16◦ (the protraction angle after contact was set to 21◦ , dotted: 0.62 Hz,
dotdashed: 0.81 Hz, solid: 1.07 Hz sweeping frequency).
distance — see Fig. 7. As expected, a shorter contact distance determined the peaks of deflection amplitude and velocity. The
produces a larger deflection amplitude. The deflection velocity, peak of deflection amplitude is found at the end of protraction.
which is calculated as a derivative of deflection amplitude, is The motor control circuit for the DC motor can adjust the
another cue to determine the contact distance. The deflection sweeping frequency and protraction/retraction angle of the
amplitude and speed in Fig. 7 are inversely proportional to the whisker. The protraction angle after the onset of contact and
contact distance, as predicted in the theoretical estimation. the total sweeping angle of the whisker were set to about 21◦
To investigate the influence of lateral shapes on deflection and 51◦ , respectively. Fig. 8 displays the average over the peaks
angles, we measured the time course of deflection signals for of deflection amplitude and velocity for 16 whisking cycles for
two different objects, a square object and a round object, and each distance of the initial contact position (140–230 mm with
D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243 235
Fig. 9. Deflection amplitude vs. protraction angle (a) square object (b) round object with diameter 110 mm (arrow: protraction angle 10◦ after contact, the deflection
amplitudes for the two different types of objects are indistinguishable at this point).
10 mm steps). The error bars show 99% confidence intervals by distance is estimated at small protraction angles and the
assuming a t-distribution. The deflection amplitudes for contact discrimination test between surface contact and point contact
distances of 180 mm or less are easily distinguishable. The is made at large protraction angles. In Fig. 9, we plot the
deflection velocity also differentiates the contact distance. For deflection amplitude depending on the protraction angles.
point contact with a square object, the deflection amplitude The deflection signal from zero up to approximately 40◦ of
and velocity have a relatively small magnitude, although they protraction angle is almost zero in amplitude, which indicates
can discriminate contact distances — see Fig. 8(a). When no deflection at all. For protraction angles larger than 40◦ , the
an artificial whisker sweeps past a round object such as a deflection amplitude increases in proportion to the protraction
cylinder, the contact distance constantly changes along the angle. Thus, we can estimate the angular position of a target
curved surface. This is reflected in the deflection amplitude and object as well as the contact distance p in the curve of
velocity as shown in Fig. 8(b). For a square object placed with a deflection amplitude vs. protraction angle. For large protraction
side angle (see Fig. 3) β = 16◦ relative to the reference axis, the angles, significant differences are visible for varying contact
whisker touched two edge-points consecutively. Fig. 8(c) shows distances and also for lateral shapes. The deflection amplitude
more pronounced deflection amplitude and velocity, since the at protraction angle 50◦ , which has a protraction of 10◦ after
contact distance is rapidly decreasing when the whisker touches contact, may provide a better estimate of contact distance,
the second edge point. The experiments imply that the type of because deflection signals at small protraction angles are less
a target object influences the deflection amplitude. If the type strongly influenced by the shape of a target object. In this
and distance of a target object are unknown, the peak values way, a set of deflection signals for small protraction angles can
of deflection amplitude or velocity may not be sufficient to determine the contact distance. Once the distance information is
localize the object precisely, or recognize the shape. Thus, we given, we can distinguish based on large protractions between a
need a systematic approach to solve the problem. surface contact and a point contact. This method only needs the
time course of deflection for a single whisker, but the distance
3.2. Lateral shape estimation at small protraction angles may have a precision
problem, when we consider the influence of noise or natural
Recognizing the lateral shape of a target object is closely frequencies on deflection signals.
related to distance estimation. Discriminating surface contact
and point contact often needs prior information of contact 3.2.2. Observing deflection amplitudes of multiple whiskers
distance at the onset of contact. Here, we will explore several When multiple whiskers in the same horizontal plane contact
approaches to estimate the contact distance or to determine the a target object, the advanced-touch whisker will have a large
contact property. variation in the deflection amplitude depending on the shape of
For the experiments, we placed a round or square object at an a target object when contact distance is fixed. In contrast, the
angular position of about 40◦ in the sweeping area of whiskers, whisker with a delayed touch has almost the same deflection
and collected deflection amplitudes of whiskers in the same level for a given contact distance over different types of
horizontal plane. objects since it has a relatively small protraction angle after
contact. Thus, for a given contact distance, the difference of the
3.2.1. Observing deflection amplitudes of a single whisker deflection amplitudes or distance estimation between the two
A set of deflection amplitudes for a sequence of protraction whiskers can discriminate point contact and surface contact.
angles is observed with a single whisker sweep. The contact Surface contact has a larger difference of deflection between
236 D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243
Fig. 10. Deflection amplitude of whisker-1 and whisker-2 (a) square object with point contact (b) round object with diameter 110 mm (arrows indicate the contact
onset moment of the delayed whisker (whisker-2), upward: distance 120 mm, downward: distance 200 mm).
the whiskers than point contact. This method will be useful affected by the vibration in their natural frequencies,3 which
if the time course of motor position or protraction angle is may cause a variation of contact onset time in the measurement.
not available. Fig. 10 shows the X –Y plot of the two whisker A more refined sensor design including damping material could
signals over time. We will attend to the moment that the improve the result.
delayed-touch whisker starts to contact the object, and take the So far we introduced methods for object localization and
moment as a reference time for the measurement of deflection lateral shape recognition. Now we describe the experiments to
angles of the advanced whisker. Arrows in the figure show the identify square or round objects using the above methods.
onset time of contact of the delayed whisker (whisker-2). The
deflection amplitude of the advanced whisker (whisker-1) at the 3.2.3. Square object
reference time is inversely proportional to the contact distance. For an edged-surface object, the deflection amplitudes
We can see in Fig. 10 that there are several branching curves depend on how the edged surface is placed in the sweeping area
which rise at different deflection angles from the horizontal of the whisker. We adjusted the angle of a square object relative
base line. To discriminate surface contact from point contact, to the axis from the rotational axis to the contact position, which
we can use the relative slope of deflection amplitudes of the two is denoted as β in Fig. 3. Fig. 11 shows an example of the effect
whiskers. A round object produces a lower slope in the curve of of varying β’s. For β’s smaller than 30◦ , the whisker touches
whisker-2 vs. whisker-1 than a square object with point contact; two-edge points and the contact distance for the second edge
the rate of the amplitude increase of the advanced whisker is point increases as β increases. Thus, small β’s result in high
higher for a surface contact than that for a point contact, while deflection amplitudes. For large β’s, for instance, β = 30◦
the delayed whisker has a similar rate of the amplitude change or greater, the whisker experiences a point contact, because it
for the two types of objects. Therefore, the X –Y plot of the two touches only the first edge point. The second whisker, which
whisker signals includes two kinds of information, the contact has a delayed touch, has almost constant deflection amplitude
with varying β’s, because it has a point contact with the first
distance and the shape of a target object. At least the plot
edge point. The graph of deflection amplitudes over the two
provides a cue to determine if the whisker has point contact
whiskers can determine the rotation angle of the square object
or surface contact.
relative to the reference axis for a given contact distance. We
We can consider another approach with multiple whiskers. can use the methods described above to estimate the contact
The time-shift or angular difference of the contact onsets distance.
between a pair of whiskers can be examined from the whisker Generally we can easily discriminate point contact and
signals and it can determine the contact distance; larger time- surface contact by observing deflection amplitude or velocity.
shift indicates closer distance. The contact discrimination test How can we determine if a square object is rotated? If the
may be done based on the contact distance. The robot knows surface slope of the object is large, it is not possible to estimate
what deflection levels would be expected at each protraction
for point contact, and can compare the measured signal with
the expected signal; if the measured signal is larger than 3 The distribution of natural frequencies depends on the boundary conditions
expected, it can be regarded as a surface contact. However, of the whisker beam. Conspicuous natural frequencies observed in the whisker
sensor signal correspond to the fixed-free boundary condition in which one tip
in the experiments, the distance estimation with the time-shift of the whisker shaft is clamped at the whisker base and the other tip is free in
often yielded a very low precision result. The whiskers are space.
D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243 237
Fig. 11. Deflection amplitudes with a square object (side length 40 mm and contact distance 160 mm) (a) whisker-1 deflection vs. protraction angle (b) deflection
curve for whisker-1 vs. whisker-2.
the rotated angle with a single whisker. It will have a point 3.2.4. Round object
contact with only one fixed contact position. If a higher level Now we investigate the deflection signal of a single whisker
of deflection is observed than expected with point contact, we for the case of round objects. Unlike edged-surface objects,
can conclude that it is a surface contact. In the time course round objects always have surface contact, and also the
of deflection (or the curve of deflection vs. protraction), we curvature of objects influences the deflection angles as shown
can find the protraction angle where there is a transition from in Fig. 5. Objects with larger curvature have a shorter contact
single-edge to two-edge point contact. In Fig. 11(a), there are distance for a given protraction angle, and as a result, they have
many branches in the upward direction from the curve for higher deflection amplitudes in time course.
β = 30◦ . The branches start at different protraction angles. The For round objects with varying curvature, we measured the
curve for β = 30◦ corresponds to point contact with a square deflection amplitudes with respect to the protraction angle in
object, and the branches with smaller β’s reflect a contact with the artificial whisker system. Fig. 12(a), (b) show examples
of the deflection amplitudes which are averaged over five
the second edge point. That is, the branching point indicates the
sweeps of whisking. For a given contact distance, a round
moment when the whisker touches the second edge point. Also
object with a larger diameter size produces a significantly
we can observe the deflection amplitude at other protraction
higher deflection amplitude than a round object with smaller
angles after the branching angle. This will determine the contact diameter. Fig. 12(c), (d) show the effect of varying curvature
distance of the second edge point. From the angular information of round objects on deflection amplitudes for a fixed contact
and the contact distance, we can estimate the side length of distance. Round objects with large diameter do not only
a square object as well as the rotated angle. For example, produce larger amplitudes, but also experience more variation
β = 12◦ has a branching point at around 11◦ of protraction after of contact distances along the surface. The difference of
contact. By the geometrical relation in Fig. 3, we can first find deflection amplitudes over varying size of round objects is more
α = 4.2◦ . Using Eq. (5), we can estimate θd = 11.9◦ , which is prominent with large protraction angles. At small protraction
very close to the rotated angle 12◦ of the square object. If we angles up to 10◦ after contact, the deflection amplitudes for
calculate the difference between the deflection amplitude for various round objects are very similar and the influence of the
β = 12◦ and the amplitude for β = 30◦ at a specific reference curvature is minute. This provides further support for using
protraction, we can obtain a pure deflection amplitude that the small protraction angles to estimate the distance for the initial
second edge point with β = 12◦ contributes to. The branching contact position.
curve has a deflection amplitude of 1 V at a protraction angle The above deflection experiments reveal that a single
of 10◦ after the branch angle of protraction, while β = 30◦ whisker, within the sensor resolution, has the potential of
has 0.6 V at the protraction angle. Thus, the second edge point distinguishing arbitrary pairs of contact distance and diameter
has a contribution of about 0.4 V at a protraction angle of 10◦ size for round objects. The robot can estimate the contact
after the contact, which corresponds to the contact distance distance of a round object at small protraction angles and
120 mm. Assuming the contact distance of the first edge point determine the diameter at large protraction angles, for instance,
at 20◦ of protraction after contact.
is estimated as 160 mm (by observing the deflection signals
of the two whiskers), we can determine the side length of a 3.2.5. Discriminating square and round objects
square object by (160 − 120)/ cos(12◦ ) = 40.9 mm, which How can we discriminate a round object from a square object
approximates the original size of 40 mm. with two edge points when their deflection curve is similar?
238 D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243
Fig. 12. Deflection amplitude vs. protraction angle for varying contact distances and varying size of round objects (a) circular disc with diameter 45 mm (b) circular
disc with diameter 100 mm (c) varying diameters of circular discs with contact distance 140 mm (D: diameter size) (d) varying diameters of circular discs with
contact distance 120 mm.
It would be a difficult task to discriminate a curved surface object at a distance of 140 mm and thus the shape information
and an edged surface including more than one point contact cannot be retrieved although we may determine if the touch
using the deflection amplitude alone, if the size and distance is a point contact or surface contact. If the robot changes the
of objects are unknown. The whiskers may not discriminate the contact distance, the response curves of deflection for the two
two objects easily, if they have a similar transition of contact objects are completely different. Thus, a collection of deflection
distances along the surface. The task needs high precision signals at different distances can provide the appropriate shape
sensors to measure the transition of deflection. Theoretically information.
their curves of deflection angle over time are different (an edged It is an advantage of multiple whiskers over a single
surface will produce a more sudden change of deflection as well whisker that a round object and an edged-surface object can
as high deflection velocity at the transition between edges), but be discriminated more precisely. In principle, a single whisker
noisy sensor signals and the presence of the whisker beam’s can distinguish point contact from surface contact, using the
natural frequencies make these differences hard to detect. deflection signal. However, adding localization information by
An alternative solution to the problem is that we can change another whisker will provide more reliable information of the
the robot position and collect another set of deflection signals. contact distance and shape of a target object.
For instance, after further moving the robot towards the target
object by 10 mm, we can again collect the deflection signals for 3.3. Vertical shape by the slip of a single whisker
the two whiskers. Then we can estimate the size and shape of
the object by observing the sequence of X –Y plot of the whisker Kaneko et al. [14] provides a geometrical analysis on
signals. Fig. 13 shows the X –Y plot of two whiskers in the same the relationship between varying force directions and the
horizontal plane for varying distances. A square object at a corresponding slip. In their experiments, the force direction of
distance of 150 mm has very similar deflection curve to a round the beam was iteratively changed towards a normal direction
D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243 239
Fig. 14. Slope test configuration; the angle of aluminium bar is adjusted for
the slope test, and the slope angle is measured counter-clockwise from the
horizontal line.
Fig. 15. X –Y plot of deflection signals (a) examples of X –Y plot over two channels with several slope tests (thick lines: estimated slope) (b) estimation of slope
with X –Y channels (◦: real data, solid line: theory)
Fig. 16. X –Y plot of deflection signals for two types of conically-shaped objects (dotted: 16◦ , dotdashed: 12◦ , dashed: 8◦ , solid: 4◦ of contact protraction angle).
Fig. 17. Time course of deflection signals for a conically-shaped object (a) deflection amplitude (b) deflection velocity (the upper and lower whiskers contact
circular disks with 105 mm and 70 mm diameters, respectively).
D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243 241
Fig. 18. Time difference and amplitude difference of two whiskers stacked vertically for relative size difference (sweeping frequency 0.77 Hz is used and the error
bar indicates 95% confidence interval by t statistic over five whisking cycles) (a) time difference (b) difference of deflection amplitude at the end of protraction
(lower whisker – upper whisker).
110 mm and the upper one has varying sizes from 25 mm or vibration speed of rodent whiskers than to deflection
to 115 mm. The discs are positioned such that both vertically amplitude [23,2,3].
stacked whiskers can touch the discs. Thus, the maximum size
difference will be 85 mm. Fig. 18 shows the averaged time 4. Discussion
difference and amplitude difference over five whisking cycles
We suggested that localization information of a target object
for varying size differences; low-pass filtered signals are used
should be the basis of shape recognition. Active whisking
as in the above experiments. The time difference was calculated
allows one to determine the distance and angular position of
by measuring the instantaneous times at the peaks of deflection
an object only by measuring the time course of deflection
velocities, where the time for the maximal deflection velocity
amplitude. The method is not affected by the elastic property
is close to the contact onset5 time. The amplitude difference
of the whisker material or by the diameter of the whisker.
was calculated by measuring deflection amplitudes at the end
The deflection amplitude and deflection velocity signals shown
of protraction. The time difference or amplitude difference is
in the experiments correspond to the activation of slowly
almost linearly proportional to the size difference of circular
adapting (SA) and rapidly adapting (RA) mechanoreceptors at
discs. As shown in Fig. 18, the amplitude difference can predict
the follicle, respectively. The two types of neurons are very
the relative size difference more exactly, since the time shift of
common in biological tactile sensors such as the skin or the
deflection between the two whiskers tends to be influenced by
whisker follicle of animals [12,24]. The former type of neurons
noise.
can signal stimulus magnitude for a given pressure, while the
As an alternative sensor technology, an array of piezoelectric
latter type of neurons cease firing in response to sustained
sensors directly produces the deflection velocity instead of
stimulation but is sensitive only when the stimulus amplitude
the deflection amplitude. A stack of piezo-type whiskers
changes. Recent physiological experiments show that SA and
provide the time-shift in the velocity signals depending on
RA neurons in the trigeminal ganglion distinctively respond to
the vertical shape of an object. Thus, the shape of an object
several modes of active whisking such as touching an object,
can be determined from the temporal whisker signal obtained
whisker bending, and detaching from an object [24]. Moreover,
for a given sweeping speed. The pattern of the velocity
it was reported that neurons in the barrel cortex encode the
signals from piezoelectric sensors is quite close to that from
deflection velocity of rodent whiskers [23,2,3], which is closely
magnetic sensors (not shown in this paper). The velocity
related with the property of RA cells. In contrast, the SA
signal of piezoelectric sensors can be a dominant feature to
cells at the follicle can differentiate the deflection level of
detect the vertical shape of an object, as is the deflection
the whisker shaft [9] and achieve directional selectivity which
amplitude measured with magnetic sensors. Unlike the above
seems to be passed to the thalamic neurons in the brain [17].
results with magnetic sensors for the vertical shape, it seems
Thus, our active whisker system is closely in line with the
that real rodents might have a preference to use deflection
biological process to use deflection amplitude and velocity
velocities rather than deflection amplitudes, because neurons
as sensory afferents. It was also shown that modeling the
in barrel cortex are more responsive to the deflection velocity
mechanoreceptors as strain sensors for deflection can produce
similar patterns of ganglion neuron firings [19]. Thus, the
5 Actually, the time for the maximal velocity does not exactly match the onset suggested method of tactile perception appears to be a plausible
time of contact, since we use low-pass filtered outputs over noisy sensor signals. mechanism for rodent whisking behaviour. The deflection
242 D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243
Table 1
Methods with artificial whiskers ; each method is described in the following sections, s1: Section 3.1, s2, s3: Section 3.2.1, s4: Section 3.3, m1, m2: Section 3.2.2,
m3: Sections 3.2.2–3.2.4, m4: Section 3.2.5, m5, m6: Section 3.4
Single whisker
Multiple whiskers
signals of whiskers can be one of the main sources of shape or shape detection in rodents. It is still an open question whether
information. rodents can discriminate complex shapes with a single whisker,
Shape information of an object is divided into two parts, or whether they interpret each whisker sensor as a simplified
lateral and vertical shape information with respect to the binary sensor or low-precision sensor and use multiple whiskers
whisking direction. Multiple whiskers provide more reliable for shape recognition processing.
information of shape or distance of a target object than
a single whisker. In addition, the shape recognition may 5. Conclusion
be involved with the integration of motion direction across
In this paper, we suggest a new approach to recognizing the
multiple vibrissae [1]. Brecht et al. [5] assumed that the
shape of a target object with a mechanical probe. The method
whiskers of rodents may be binary sensors which only tell
uses biomimetic whiskers to model the design of the follicle
if touched or not, and that the clusters of whiskers at each
of rodent whiskers. Deflection amplitude or deflection velocity
side of face play a major role in detecting the distance of
signals provide localization information of a target object, that
an object. Our experiments support that each whisker has the
is, contact distance and angular position of an object. Also they
capability to provide continuous distance information rather show the potential of detecting lateral shape and the vertical
than just binary information as well as to track directional cues. shape of objects. For lateral shape recognition, we found the
The shape can be derived from a combination of distance, the deflection amplitude changes depending on whether a sweep
two channel information of a whisker, or the time difference of whiskers experiences point contact or surface contact along
of contact onsets retrieved using multiple whiskers. It seems the surface of a target object. This effect is caused by the
that head movement is involved with shape recognition or dependency of deflection on the contact distance. The deflection
texture discrimination of rodents [5,11]. In our experiments, signal helps discriminate round objects with a varying curvature
precise recognition of shape, for example, to distinguish a or objects with different lateral shapes, like round and square
curved surface from an edged surface, would need a collection objects. We present several methods to estimate the contact
of whisking signals at different robot positions, which may distance and to detect the lateral shape of a target object with a
resemble the behaviour pattern that rodents use to refine tactile single whisker or with multiple whiskers. Multiple whiskers can
sensory information. We summarize the methods used for provide more precise information of shape and contact distance
distance estimation and shape recognition with our artificial rather than a single whisker.
whiskers in Table 1. A single whisker with two channels arranged orthogonally
In principle, the time course of deflection signals with to each other can recognize the slope of the vertical shape.
a single whisker can be a direct source of lateral shape Alternatively, the vertical shape can be estimated by two or
information, if high precision of sensor signals are available. more whiskers stacked vertically. For instance, a conically
Our experiments show that a single whisker alone can shaped object will produce a time-shift in whisker signals
distinguish point contact and surface contact, but it may have between the upper whisker and the lower whisker. The
difficulty in discriminating a curved surface and an edged time-shift of contact can be measured by observing the
surface, since the precision of the deflection signal is reduced by deflection velocity. Another approach is to use the time course
noise and natural frequencies. So far there is no physiological or of deflection amplitudes, since the difference of deflection
behavioral data available on the precision of tactile localization amplitudes also provides the vertical shape information.
D. Kim, R. Möller / Robotics and Autonomous Systems 55 (2007) 229–243 243
The suggested methods for shape recognition may provide [18] M. Lungarella, V.V. Hafner, R. Pfeifer, H. Yokoi, An artificial whisker
a clue to how real rodents process tactile signals of their sensor for robotics, in: IEEE Int. Conference on Intelligent Robots and
whiskers for shape recognition, and can be applied to provide Systems, IEEE, 2002, pp. 2931–2936.
[19] B. Mitchinson, K. Gurney, P. Redgrave, C. Melhuish, M. Pearson,
autonomous robots with the same capabilities. I. Gilhespy, T. Prescott, Empirically inspired simulated electro-
mechanical model of the rat mystacial follicle-sinus-complex, Pro-
Acknowledgement ceedings of Royal Society B: Biological Sciences 271 (1556) (2004)
2509–2516.
This work has been supported by the EU in the project [20] D.J. Pinto, J.C. Brumberg, D.J. Simons, Circuit dynamics and coding
AMOUSE (IST-2000-28127). strategies in rodent somatosensory cortex, Journal of Neurophysiology 83
(3) (2000) 1158–1166.
[21] R.A. Russell, Using tactile whiskers to measure surface contours,
References
in: International Conference on Robotics and Automation, IEEE, 1992,
pp. 1295–1299.
[1] M.L. Andermann, C.I. Moore, A somatotopic map of vibrissa motion
[22] A. Seth, J. McKinstry, G. Edelman, J. Krichmar, Spatiotemporal
direction within a barrel column, Nature Neuroscience 9 (2006) 543–551.
processing of whisker input supports texture discrimination by a brain-
[2] E. Arabzadeh, R.S. Petersen, M.E. Diamond, Encoding of whisker
based device, in: S. Schaal, A. Ijspeert, A. Billard, S. Vijayakumar,
vibration by rat barrel cortex neurons: Implications for texture
J. Hallam, J.-A. Meyer (Eds.), From Animals to Animats 8, Proceedings
discrimination, Journal of Neuroscience 27 (2003) 9146–9154.
of the International Conference on the Simulation of Adaptive Behavior,
[3] E. Arabzadeh, E. Zorzin, M.E. Diamond, Neuronal encoding of texture in
MIT Press, 2004, pp. 130–139.
the whisker sensory pathway, PLOS Biology 3 (1) (2005) 155–165.
[23] M. Shoykhet, D. Doherty, D.J. Simons, Coding of deflection velocity
[4] S. Bovet, R. Pfeifer, Emergence of delayed reward learning from
and amplitude by whisker primary afferent neurons: Implications for
sensorimotor coordination, in: Proceedings of the Int. Conf. on Intelligent
higher level processing, Somatosensory and Motor Research 17 (2) (2000)
Robots and Systems, IEEE, 2005, pp. 841–846.
171–180.
[5] M. Brecht, B. Preilowski, M.M. Merzenich, Functional architecture of the
[24] M. Szwed, K. Bagdasarian, E. Ahissar, Encoding of vibrissal active touch,
mystacial vibrissae, Behavioural Brain Research 84 (1–2) (1997) 81–97.
Neuron 40 (2003) 621–630.
[6] G.E. Carvell, D.J. Simons, Biometric analyses of vibrissal tactile
[25] T. Tsujimura, T. Yabuta, Object detection by tactile sensing method
discrimination in the rat, The Journal of Neuroscience 10 (8) (1990)
employing force/torque information, IEEE Transactions on Robotics and
2638–2648.
Automation 5 (4) (1989) 444–450.
[7] M. Fend, Whisker-based texture discrimination on a mobile robot,
[26] N. Ueno, M. Kaneko, Dynamic active antenna, in: Proc. of Int. Conf. on
in: Proceedings of the European Conf. on Artificial Life, Springer Verlag,
Robotics and Automation, IEEE, 1994, pp. 1784–1790.
2005, pp. 302–311.
[27] J.F. Wilson, A. Chen, A whisker probe system for shape perception of
[8] M. Fend, S. Bovet, H. Yokoi, R. Pfeifer, An active artificial whisker array
for texture discrimination, in: IEEE Conf. on Int. Robots & Systems, solids, ASME Journal of Dynamic Systems, Measurement and Control
IEEE, 2003, pp. 1044–1049. 117 (1995) 104–108.
[9] K.-M. Gottschaldt, A. Iggo, D.W. Young, Functional characteristics of
mechanoreceptors in sinus hair follicles of the cat, Journal of Physiology DaeEun Kim received his B.E. and MSc(Eng) in
235 (1973) 287–315. the department of computer science and engineering
[10] J.A. Hartings, S. Temereanca, D.J. Simons, High responsiveness and from Seoul National University and the University
direction sensitivity of neurons in the rat thalamic reticular nucleus to of Michigan at Ann Arbor, respectively. He was a
vibrissa deflection, Journal of Neurophysiology 83 (5) (2000) 2791–2801. lecturer in Korea Air Force Academy from 1994 to
[11] M.J. Hartmann, Active sensing capabilities of the rat whisker system, 1997 and then joined a research team involving with
Autonomous Robots 11 (2001) 249–254. the bus-underground smartcard ticketing system in
Seoul. He received Ph.D. degree from the University
[12] E.R. Kandel, J.H. Schwartz, T.M. Jessell, Principles of Neural Science,
of Edinburgh in 2002. From 2002 to 2006, he was
fourth ed., McGraw-Hill, New York, 2000.
a research scientist in Max Planck Institute for Human Cognitive and Brain
[13] M. Kaneko, N. Kanayama, T. Tsuji, Vision based active antenna,
Sciences. Currently he is a senior researcher in University of Leicester. His
in: International Conference on Robotics and Automation, IEEE, 1996,
research interests are in the area of biorobotics, autonomous robots, artificial
pp. 2555–2560. life, neural networks and neurobiology.
[14] M. Kaneko, N. Kanayama, T. Tsuji, Active antenna for contact sensing,
IEEE Transactions on Robotics and Automation 14 (2) (1998) 278–291.
[15] A.G. Kelly, Fundamentals of Mechanical Vibration, second ed., Mc-Graw Ralf Möller received a Ph.D. in electrical engineering
Hill, 2000. from the Technical University of Ilmenau, Germany,
[16] D. Kim, R. Möller, A biomimetic whisker for texture discrimination and a Venia legendi in computer science from the
and distance estimation, in: S. Schaal, et al. (Eds.), From Animals to University of Zurich, Switzerland. He is heading
Animats 8, Proceedings of the International Conference on the Simulation the Computer Engineering Group at the Faculty of
of Adaptive Behavior, MIT Press, 2004, pp. 140–149. Technology of Bielefeld University. His research
[17] S.H. Lichtenstein, G.E. Carvell, D.J. Simons, Responses of rat trigeminal interests include neural networks, models of visual
ganglion neurons to movement of vibrissae in different directions, cognition, visual robot navigation, and biorobotics.
Somatosensory and Motor Research 7 (1) (1990) 47–65.