Table 1: Summary of validation results.
RMS Error 1.6 mm
Max Error 4.6 mm
Calculation Time <1 second
Carmine 1.09”. The recorded data can be seen in
Figure.7 where as the model prediction can be seen in
Figure 8. The visualization shown in Figure 8 takes
less than one second to calculate using non optimi-
zed MATLAB code on a conventional PC. The metric
used for this evaluation only takes the offset in the ”z”
direction into account, and does not perform a one to
one mapping of points in the ply. Seen over the whole
test area (5 by 3 suction cups), the maximum devia-
tion between ply and model is 4.6 mm and the RMS
error is 1.6 mm. This is comparable to the spatial re-
solution of 1 mm for the used camera configuration.
This is already quite satisfying for the purpose of le-
arning draping strategies and with better suction cups
and camera estimates, we expect the accuracy to be
further improved.
It should however be noted that there in the test
data are some unrealistic discrepancies with respect to
suction cup positions. These test data errors give rise
to the models arc length adjustment failing on some
lines, and as a result some unexpected bulges are for-
med.
7 CONCLUSIONS AND FUTURE
WORK
In this paper, we have presented a fast computable
model for predicting the shape of prepreg fiber plies
clamped in the corners by suction cups. Our presen-
tation included a detailed derivation of the model pa-
rameters as function of the suction cup positions and
orientations. We also did some preliminary experi-
mental validations and found that the accuracy of the
model is promising. However, the quality of the expe-
rimental setting was partly insufficient and therefore
further validation in is needed.
Together with our partners, we are currently im-
proving the experimental settings by replacing the
suction cups with new versions customized for com-
posites. Furthermore, the sensor settings is being sig-
nificantly improved. Our model currently only sup-
ports completely "free hanging" regions of the ply. In
the near future, we will therefore extend the model to
include situations where part of the region between
four suction cups has made contact with the mould
and where we estimate only the shape of the remai-
ning part. This will lead to some additional issues
because we need to estimate the boundary of the free
hanging part, which will be a 3D curve and interpo-
late the shape from that boundary to the remaining 1,
2 or 3 suction cups.
ACKNOWLEDGEMENTS
This work was supported by Innovation Fund Den-
mark through the strategic platform MADE - Platform
for Future Production and the project FlexDraper.
REFERENCES
Do, D., John, S., and Herszberg, I. (2006). 3d deformation
models for the automated manufacture of composite
components. Composites Part A: Applied Science and
Manufacturing, 37(9):1377–1389.
Ellekilde, L., Wilm, J., Nielsen, O. W., Krogh, C., Glud,
J. A., Gunnarsson, G. G., Stenvang, T. S., Kristian-
sen, E., J. Jakobsen, M. K., AanÃ˛es, H., de Kruijk,
J., Sveidahl, I., Ikram, A., and Petersen, H. (2018).
An outline of the design and control of the flexdraper
robot system for automatic draping of prepreg compo-
sites fabrics. To be submitted.
Kot, M., Nagahashi, H., and Szymczak, P. (2014). Elas-
tic moduli of simple mass spring models. The Visual
Computer, pages 1–12.
Krogh, C., Glud, J. A., and Jakobsen, J. (2017). Modeling of
prepregs during automated draping sequences. In AIP
Conference Proceedings, volume 1896, pages 030–
036. AIP Publishing.
Lin, H., Ramgulam, R., Arshad, H., Clifford, M., Potluri, P.,
and Long, A. (2012). Multi-scale integrated modelling
for high performance flexible materials. Computatio-
nal Materials Science, 65:276–286.
Liu, T., Bargteil, A. W., O’Brien, J. F., and Kavan, L.
(2013). Fast simulation of mass-spring systems. ACM
Transactions on Graphics (TOG), 32(6):214.
Newell, G. and Khodabandehloo, K. (1995). Modelling
flexible sheets for automatic handling and lay-up of
composite components. Proceedings of the Institu-
tion of Mechanical Engineers, Part B: Journal of En-
gineering Manufacture, 209(6):423–432.
Zhou, C., Jin, X., Wang, C. C., and Feng, J. (2008). Plau-
sible cloth animation using dynamic bending model.
Progress in Natural Science, 18(7):879–885.