the clay in the backward phase, there is a delay between
the torso and arms. The motion consists of two parts
with phase relationship, forming a single cycle with two
modes. We think that this phase relationship is acquired
at the final stage of skill acquisition, following the estab-
lishment of coordination. We discuss below this issue
based on our experimental results.
Figure 2: trajectories of kneading (subject A) on the sagittal
plane
right knee
left hip
right hip
30 30.5 31 31.5 32 32.5 33 33.5 34
Time[s]
Figure 3: time series of subject A (expert)
Figure 4: time series of subject B (experienced)
4.2 differentiation within coordination in
kneading
As previously reported (Abe et al., 2003), we found the
trajectories to be well coordinated as a person gets more
experienced in the task. While only a qualitative anal-
ysis was presented in our previous work, we present in
this paper a quantitative analysis of the coordination.
Let us begin with a description of coordination. In our
experiment, the localisation of trajectories are found for
both the experienced persons and experts (i.e., subjects
A, B, C and D). Although we previously found an ex-
pert’s trajectory to be highly localised, the experts who
took part in the current experiment did not show such
strong localisation. Contrarily, the experienced persons
(subject B and C) showed better localisation as long as
trajectories in the real space is concerned.
Figures 3 and 4 depict the time series of subject A and
B, respectively. The latter may give better impression
of strong synchronisation. Applying FFT (Fast Fourier
Transformation), we found that subject B’s motion has
sharper peaks for all angles while subject A’s has a single
broader peak. Then if we only consider the strength of
synchronisation, subject A is regarded to be inferior to
B.
Subject A is, however, more skillful than the others as
long as we judge their skills based on their end products.
It is important to note that the establishment of coordi-
nation is merely halfway to skill acquisition as described
below.
Next, we turn to analysis of differentiation. In the
present work, we evaluated the relative phases among
joints and confirmed our previous findings.
In Figures 5 to 7, the histograms of relative phase are
shown for subjects A, B and C, respectively. The refer-
ence angle was chosen to be the hip of anterior stance
side, i.e., left hip. Data are extracted from a single trial,
but qualitative features are preserved.
For all the subjects except E (i.e., novice), a synchro-
nisation is observed between the reference angle and the
lumbar. The other hip is almost synchronising although
a small delay was found in subject A.
In the torso, we do not find any synchronisation of the
neck. This may be due to that reduction of our body
model (i.e., the chest and top of the neck). Especially,
the movement of viewpoint affects the attitude of the
head (i.e., top of the neck) and we expect a coordination
to be observed if we have adopted a more detailed body
model.
The movement of arm is most important part of the
motion because only the hands touch the clay physically.
As noted, one possibility for the efficient way of knead-
ing is to use the gravitational force to help to press the
clay down. This pattern is found among the experienced
persons. In Figures 6 and 7, the arm is synchronising
(subject C) or proceeding (sub ject B) to the hip. They
tended to push the clay with the help of momentum of
the whole body. By stretching the elbow prior to swing
down of the body, the arms become stiff and easily press
the clay to stretch it out. A disadvantage of this move-
ment is that the rest of a cycle is only to swing back the
100