Sep 19 2017

Soft Robots

vaderhandPart of the reason I like science fiction is because it can be a thought-experiment about future technology and society. For this reason, like any self-respecting nerd, I often pay close attention to the details of how future technology is portrayed. Stepping out of the movie-as-entertainment and storytelling for minute, and focusing on the ideas that went into crafting a vision of technology advanced beyond our current tech – how did they do? What assumptions did the creators make, and did they break any laws of physics.

I tend to be most fascinated by what hidden assumptions and biases go into visions of the future. I wrote about this previously with respect to spaceship design. The same idea applies to how robots are portrayed in science fiction.

For example, perhaps my biggest peeve is Star Wars (probably because it is a beloved franchise). I don’t necessarily mind the design of droids, because you can argue that they are intended to look unmistakably mechanical. However, we get a few glimpses of artificial robotic limbs – Luke’s hand and Vader’s arm, for example. These appear to be entirely constructed of wires and pulleys. Vader’s arm is the worst – the stump is nothing but a mass of wires.

This image of artificial robotic parts is almost ubiquitous in science fiction, although recently there are some great examples of moving past this cliche. Just compare the original West World, where robot faces look convincingly human, until they are removed and revealed to be rigid metal with wires underneath. The recent series West World, however, portrays soft 3D printed robotic technology. That’s what I’m talking about. Soft Robots

There are many reasons why we would want to develop soft robotic technology. In industry there are applications that require flexibility, including manipulating small or soft parts. We also might want soft robots anytime they need to interact with humans or other living things, especially medical applications.

Soft life-like prosthetics are the holy grail. We can make soft skin, flexible joints, and a rigid frame. Perhaps the biggest obstacle at this point is making soft artificial muscle. We are making some progress, which I will get to shortly, but it remains difficult to imitate living muscle.

Like many technologies I discuss here, in order for artificial muscles to be useful they need to have a suite of simultaneous properties, the lack of any one can be a deal-killer. The ideal battery, for example, needs to have high capacity, low weight, many recharge cycles, relatively quick recharging, able to discharge fast enough to power demanding devices, should be stable (not burst into flames), and be made of materials that are cheap, abundant, and non-toxic.

For artificial muscles they need to be soft, have high tensile strength, and be able to contract and expand repeatedly without limit, with both rapid and fine control, able to produce significant power, under conditions compatible with a living user. A technology can have most of these properties but be very slow, and therefore useless as a prosthetic, and only potentially useful for narrow niche applications. Another technology might be too weak to be of practical use, even if everything else is perfect.

As with battery and other technologies, I often read about “breakthroughs” where one feature is significantly advanced, and the press release glosses over the glaring deficits or casually states, “Researchers are now working on fixing this deal-breaking deficit, which they are confident they can do in the next funding cycle.”

This brings us to the latest news on soft robotics which prompted this post – researchers at Columbia University have designed a soft artificial muscle that is able to contract and expand with electrical or thermal control, and:

 The new material has a strain density (expansion per gram) that is 15 times larger than natural muscle, and can lift 1000 times its own weight.

Impressive – but where is the deal breaker? Here’s one:

After being 3D-printed into the desired shape, the artificial muscle was electrically actuated using a thin resistive wire and low-power (8V). It was tested in a variety of robotic applications where it showed significant expansion-contraction ability, being capable of expansion up to 900% when electrically heated to 80°C. Via computer controls, the autonomous unit is capable of performing motion tasks in almost any design.

Requiring heat of 80°C (176°F) is not practical for human applications. This could be useful in industrial applications, however. Also, it seems like it can be actuated just with electrical stimulation of 8 volts, which could work. But here is the real limitation:

The researchers will continue to build on this development, incorporating conductive materials to replace the embedded wire, accelerating the muscle’s response time and increasing its shelf life.

The “response time” is the critical bit. Watch the video’s on the link – the contraction and relaxation of the artificial muscle is slow. It literally moves at about the speed of a sloth, so this would be great for robotic sloths, but not so much for human prosthetics.

As always, I hope the researchers are successful in solving these issues. The response time is a non-trivial problem, however, and may be intrinsic to the technology. Popular discussions (flowing from the press release) of such technology always assumes that the advance is on a trajectory to the final goal, but that is often not the case. We may not get to an artificial muscle with human-level or greater properties via this technology. This could easily be a dead end if the response time problem is unfixable.

Still, even a dead end can be useful if it teaches us something, and may help point to the way to whatever technology will ultimately work. Or maybe not – we still don’t have room temperature superconductors or a viable storage option for hydrogen fuel cells, despite all the advances made.

What this means is that it is difficult to predict when we will cross the finish line to a viable technology. Success is not simply a matter of tweaking or iterating current technology. It may require an entirely new material or approach, and such advances are impossible to predict. I do think we’ll get there eventually, but who knows how long it will take.

Human-level soft robotic muscles, however, would be a tremendous technology to have, and will rapidly revolutionize the prosthetic industry.



8 responses so far