Researchers from the Universities of Southern California, UC Irvine, and Bowdoin College have published a paper pointing to an unlikely inspiration for the future of simpler and more decentralized robotics systems: sea stars, also known as starfish.
"The nervous system does not process everything in the same place at the same time, but relies on the idea that the sea star is competent and will figure it out," explains USC Viterbi School of Engineering Professor Eva Kanso, one of the researchers, of what makes the sea star's tube-footed locomotion special. "If one tube foot pushes against the ground, the others will feel the force.
"This mechanical coupling is the only way in which one tube foot shares information with another. In the case of the sea star, the nervous system seems to rely on the physics of the interaction between the body and the environment to control locomotion. All of the tube feet are attached structurally to the sea star and thus, to each other."
That same technique can apply to mechanical locomotion, too, the researchers claim. "Using the example of a sea star, we can design controllers so that learning can happen hierarchically," Professor Kanso explains. "There is a decentralised component for both decision-making and for communicating to a global authority.
"This could be useful for designing control algorithms for systems with multiple actuators, where we are delegating a lot of the control to the physics of the system — mechanical coupling — versus the input or intervention of a central controller."
The team has published its work in the Journal of the Royal Society Interface, under open access terms, and is moving on to investigating what happens in a sea star-like system when there are competing stimuli.