![]()
While the current generation of industrial robots is primarily made of metal, the research community has been getting interested in the potential for soft-bodied robots. These have a number of advantages, such as being easy to customize via 3D printing and providing a flexibility that lets them squeeze through tight spaces.
Many of the research demonstrations created so far, however, have required some compromises. For some iterations, this has meant the control hardware and power sources have been kept separate, connected to the robot via a tether. For other attempts, this has meant the final product is a mixture of hard and soft pieces.
In today's issue of Nature, however, researchers are reporting the creation of a soft-bodied robot that carries its own fuel supply, which powers the robot through an on-board chemical reaction. Soft, flexible on-board logic then directs the reaction products to control the movement of the robot. While the result is pretty limited in what it can do, its creators make up for that with a certain cool factor, making their creation look a lot like an octopus.
By this point, soft 3D-printed robots have a bit of a history. Early versions were typically powered by compressed air fed into them through tubes, which caused different parts of their bodies to flex. Generation of the compressed air took place outside the robot, as did the control over which parts of the robot received it. Still, these robots demonstrated potential, as they were able to squeeze through narrow cracks by flexing carefully. A later version also allowed these robots to change color.
Other researchers figured out how to get rid of the need for an external power supply. By sacrificing a bit of flexibility, they added a small butane tank that allowed their robot to make explosive leaps when a chamber full of butane was lit. The partially rigid structure also enabled them to place the controller on board.
The new entry, which its creators are calling octobot, combines some elements from the two approaches. It carries its fuel on board, but the fuel is stored in a flexible tank. The fuel undergoes a chemical reaction on board as well, rather than in an external combustion chamber. In fact, the fuel doesn't combust at all; it's a solution of hydrogen peroxide in water, something you might find in your medicine cabinet. When in contact with a platinum catalyst (also on board), this chemical (formula: H2O2) forms water and releases oxygen. The oxygen gas is then used to power the robot's movements just as compressed air was in the earlier robot.
The authors actually had to work to avoid having combustion take place. They ended up using a solution that's 50-percent hydrogen peroxide, even though it's possible to obtain higher concentrations. The reason they used this concentration is that "concentrations above 50 percent [by weight] drastically increase the decomposition temperature, resulting in combustion within the printed catalytic reaction chambers." In other words, higher concentrations set the robot on fire while feeding it a steady supply of oxygen.
To provide controlled movement, a series of valves determines where the fuel gets sent, as different reaction chambers sent oxygen gas to different octobot arms. Once there, the gas inflated a chamber, causing the arm to flex. A valve allowed the gas to slowly drain out, allowing the arm to return to a relaxed configuration. Right now, this just raises and lowers the arms, though it's conceivable that with a slightly different geometry the arms could move octobot around.
The control mechanism is actually a simple series of valves with feedbacks. When one set of valves is open, it pinches off the fuel supply of the second. The feedback eventually flips this so that the fuel supply oscillates between feeding two reaction chambers.
Details like the size of the reaction chambers and the inflatable portion of the arms ended up having to be worked out by trial and error. Here, 3D printing turned out to be very useful, as the octobot's makers were able to test 30 overall designs and about 300 individual iterations to optimize performance.
Overall, it's an impressive effort only undercut by the final product: this octopus-shaped robot does little more than sit in place and flex alternate sets of limbs. The general approach clearly has a lot of potential, and I look forward to seeing what the research team does with some of the technology they've developed. I'd just rather see a bit of it now.
More generally, efforts like this one are building the equivalent of a robotics toolset, one that will allow us to pick and choose different technologies based on our needs. It's entirely possible that in a few years we'll see robots that are a mix of hard and soft parts, possibly with more than one power source on board.
Nature, 2016. DOI: 10.1038/nature19100 (About DOIs).
