Autonomous Embodied Evolution of a Biomimetic Robot's Rhythmic Motion Behavior
Results much truer to the body-control-dynamics
would be accomplished if the needed evaluation feedback would originate
from the robotic body itself.
Further development of the projects' core ideas would
demand a more intensive study of possible evolutionary strategies. The long
time-requirements for conducting the tests largely inhibited experimental
investigations on different strategies. More carefully designed selection
and reproduction mechanisms and better tuned setup parameters might improve
the obtained results.
Results much truer to the body-control-dynamics would be accomplished if the
needed evaluation feedback would originate from the robotic body itself. This
could be achieved by employing stretch sensors in parallel with the spring
modules or by incorporating rotation sensors into the mechanical joint design.
A revision of the robots' mechanical design would allow the introduction of
modularity into the system. The main morphology of the robot could become
subject of the evolutionary development as well.
Instead of constructing the robotic bodies out of wood and metal and just
granting the muscle elements elasticity, the whole structure could be manufactured
out of compliant materials. The possible use of shape decomposition methods
would add additional variability in the construction process. In becoming
more soft and flexible the robots would underline the merge of biology and
Success in the development of efficient and easy-to-handle EAP products might
provide a future alternative for the SMA springs. The polymers promise a greater
force-potential and could free the robotic sculptures from their untouchable
position in suspension.
(c) 2007 - Thesis project by Eva Schindling
Completed in the MSc. programme Art & Technology