New 'Rubber Robot' Crawls Through Small Spaces With Inflatable Limbs 56
JoeRobe writes "Researchers at Harvard have developed a pneumatically-controlled rubber robot that combines undulation and quadrupedal 'crawling,' allowing it to maintain a low profile while moving. In a paper published in PNAS (abstract), they describe it as a 'soft robot, composed exclusively of soft materials (elastomeric polymers), which is inspired by animals (e.g., squid, starfish, worms) that do not have hard internal skeletons.' The robot is solely powered by relatively low pressure (10 psi), and controlled by 5 pneumatic actuators. The research was funded by DARPA." The paper is also available (not paywalled) from the researchers' project site (PDF), complete with more creepy images of the squidbot.
next phase design? (Score:4, Interesting)
Ok, jokes about inflatable rubber robots crawling through small spaces aside....
I imagine a kind of pneumatic (or better, hydraulic) soft bodied robot that has a kind of "fluid logic" system that squeezes off the supply lines to the "muscle chambers", and a single internal electrical contraction chamber that serves as the main pump.
Say for instance, the main pump simply squeezes a large, central space somehow, and has say, 8 output lines from that chamber. Surrounding each line is a "pinch" cell, which is a smaller version of the main pump cell, but with no output lines. When active, it constricts the fluid flow down the tube it surrounds, preventing working fluid from passing.
The tubes continue and terminate at the expansion chambers in the robot's limbs.
The computer circuitry which drives the main cell, and the 8 control cells, is tiny and self contained. Possibly an epoxy blob design. This would be the only rigid part of the robot, though it too could be flexible with the right design. (Use flexible substrate electronics, like that graphene ink approach.)
Depending on how those 8 control valves are opened or closed, the action of the main pump cycle would actuate the robot. 8 valves would give you 256 possible muscle instructions to work with.
Conceivably, you could even do "tricks" this way, by inflating a limb motor cell, then blocking the fluid return by closing the valve, to hold one part of the robot stiff, while the remaining working fluid drives some other part of the robot, to accomplish some task. An example use might be "jumping", since the release of the control valves would violently snap the muscle as the pressure released.
I could see some pretty complex movements being done this way.
The issue is how to power the robot. A laboratory test robot might walk on a special floor which doubles as an inductive charger, which powers the robot's electronics, but a real world softbody robot would something a little more real.