(a).  This comply with the well known non-linear behaviour of SMA. The
reason is this: percentage of austenite rapidly increases with the first part
of thermal energy provided to material by Joule effect; after that, less and
less material still remains martensitic. To make this remaining part of ma-
terial transform into austenite, a thermal diffusion in the material for con-
duction has to be realised, so that all the parts of alloy to reach activation
temperature. From these tests, a characteristic force versus displacement
has been deduced, see Fig. 2 (a).This characteristics (a) is nearly linear; this confirms that in austenitic
phase material behaves as perfectly elastic, at least for little displacements,
while at larger strains a shift from linearity can be noticed. This shows that
from the point of view of actuation forces, there are no preferential dis-
placements to work with, since slope is constant as for traditional springs.
More critical for our goal is the time SMA takes to reach flex point, The
results are shown in the graph (b).  From this graph, it can be noticed that
time do not grows linearly with starting displacement, but it has an increas-ing slope. Therefore, from the point of view of speed of actuation, it is fa-
vourable to set low displacements for the springs, in order to work with
shorter times of heating for following shortening.
Using this data, it is possible to determine the needed bias spring.
The main problem using SMA for repeated actuation is long time for
SME to regain martensitic, thus plastic, state. So, generally speaking,
SMAs are unsuitable for actuation of fast devices, but speed is not a pecu-
liarity of peristaltic actuation; however, an optimal duty cycle has to be
found in order to determine how fast the robot will crawl, that is the high-
est frequency of peristaltic waves that can be applied.
From the point of view of readiness of actuation, thus in growing slope,
we  noticed that response times have a magnitude of some 0,1s. Then we
evaluated the time needed by coils to relax to their residual strains, by
heating them for a time long enough to reach complete austenitic phase,
then measuring time taken from springs to regain martensitic state.
Finally, from the results obtained within the duty cycle tests campaign,
two non-saturating cycles were found for a 60 mm elongation, which is
suitable for our robot. These were 4-20 and 3-15 cycles, with 0,166 duty
factor. These sequences proved good for their stability, in fact, force even
after various cycles stops to increase its medium value, keeping a good
amplitude that is needed for actuation.  
On the base of these evaluations, the physical prototype of the locomo-
tionn modules has been successfully built and a control system with non-
linearities compensation is defined and tested.
Conclusions and future work
The design of a peristaltic locomotion module for worm like robot has
been presented. After suitable parametric 3D modeling, computational
analyses and laboratory experiences, actuation by SMA springs disposed
in a symmetrical configuration has been chosen. The physical prototype of
a module has been built using commercial materials, this solution proved
good during tests and it has been also found how, using tailored materials
instead of commercial ones, its behaviour could still improve.
Acknowledgement
The civil protection units and fire department of Liguria region are kindly
acknowledged for the definition of specifications and requirements of our worm like robots. We thank CRF and GE for the critical discussion on the
materials choice and tests results. 摘要：本文研究的蠕动运动，主要适应于使用检查和抢救的应用。一个蠕动运动状态的艺术和有关适用的智能材料的一个简短的概述。一个有效的解决方案，运用形状记忆合金（SMA）弹性讨论。对工作原理进行说明，并提出了一个初步的原型。 蠕动运动蠕虫仿生机器人英文文献和翻译(4):http://www.youerw.com/fanyi/lunwen_3732.html