Hacker News new | past | comments | ask | show | jobs | submit login
The first autonomous, entirely soft robot (news.harvard.edu)
189 points by krnekdo on Aug 29, 2016 | hide | past | favorite | 44 comments



I love project and it's a great implementation of novel ideas all slapped into one package, but I do take some issues with how they describe it.

It might make it under a liberal definition of 'robot', but it certainly doesn't fit under what people think of as robots in a modern society. Also, it's just an oscillator in a novel implementation, not 'autonomous'. It's like saying my oscillating fan is autonomous.

Awesome progress, look forward to future development, but a little to hyped up for my tastes.


Even among experts, there is no consistent, widely-accepted definition of "robot"...


The term is actually very well defined:

>> a machine capable of carrying out a complex series of actions automatically, especially one programmable by a computer.

The Mars rovers... robots. Squishybot here is a technology platform demonstrator. It shows that you could build one with these techniques.


So... your dishwasher, printer, washing machine, oven, and thermostat? All carry out "complex actions automatically" and are controlled by computers. No one would consider these "robots".

With all due respect, there is no agreed-upon definition. (As someone with a PhD in robotics, CEO of a robotics company, & robotics journalist for 10+ years. Even my old academic advisor and I had contention around 3D printers.) If that's not enough, here's a proper source: http://robohub.org/robohub-roundtable-why-is-it-so-difficult...

The best definition I've ever heard: A robot is a device that doesn't work (yet); as soon as it works, it is renamed (Vacuum robot => Roomba; clothes cleaning robot => washing machine; etc.)


"A robot is a device that doesn't work (yet); as soon as it works, it is renamed"

I love this definition.


I might be in minority but I always considered most them robots.

(depending of course the model and complexity)

But then again, I am no one


By that definition a card board mock up would be a robot too. I agree with the grandparent that a squishy thing that just flails its arms is not what most people would consider a "robot", experts included.


> washing machine, oven, and thermostat? ... No one would consider these "robots".

> clothes cleaning robot => washing machine; etc.

So do you think a washing machine is a robot?


Nope. It already works.


But that would mean that washing machines were at one time considered robots, but now they aren't ... Ohh, okay, I like that definition too.


So, same definition as AI?


I think 'autonomous' is the more problematic term in the article because implies some kind of `intelligence`. AFAIU this `robot` doesn't do much to be considered intelligent in any way.


Perhaps it is easier to agree on what is not a robot.

I am not a robot.

Wait, am I?


What makes any robot not "just an oscillator"? Aren't CPUs just oscillators in a novel implementation? Does it have to be made of semiconductor? Metal?


While it might not be perfectly encompassing, I would generally consider a "robot" to necessarily use some form of Turing Machine executing software to control its movement.


From metafilter: "They call it the first fully autonomous soft-bodied robot, but I prefer to think of it as the first pneumatically self-propelled Wacky Wall Walker." (user 'Strange Interlude')

Actually pretty true.


It is. And when you think of it that way, why not replace the pretty unsafe [1] hydrogen peroxide with compressed air instead? Academic exercises aside, I don't see how a flow control valve is any less "soft" than having a platinum catalyst. But hey, they're chemists, and when what you have is a hammer...

[1] Wikipedia says "pure hydrogen peroxide will explode if heated to boiling, will cause serious contact burns to the skin and can set materials alight on contact." Not to mention it's been widely used for bomb manufacture by terrorists, so today trying to obtain anything stronger than 1% H2O2 as a normal person gets you on a watchlist, or even a phone call from the FBI/local equivalent.


Thing is, it doesn't even translate itself, so it's really more of a 'rocket fuel powered wacky waving tentacle.'


Great idea, though it probably comes to description of things like 'robot'.

To be more precise, from the article I understood that this is a 'soft equivalent of a simple electronic oscillator'.

So, probably, I will not call this a robot (but probably we would never have heard of this if not the click-baity title), but it's neat idea which could be worked on further.


Honestly, I am disappointed that even though they built a circuit with an oscillator they didn't make a robot that moves.

They could have easily made a manta ray robot like this: https://www.youtube.com/watch?v=mTjhXKIXc90


That's its control system. What's interesting is that all of its actuators are soft. You can imagine that a set of simple soft components like that simple oscillator, combined with some small rigid components containing a microcontroller, would belike a full soft robot.


Some definitions of 'robot' require a sensor too?


In the broadest definition, a sensor is an object whose purpose is to detect events or changes in its environment, and then provide a corresponding output. [1]

If we follow the above definition given on the Wikipedia then some important and interesting issues arise:

Is sensor not a device that behaves in certain manner to some energy stimulus and thus allows us to connect outer energy to the autonomous system (robot) being developed?

Should a sensor always transduce to electronic energy form?

e.g. a gear lever in a bike is not a sensor? it senses (may be pressure) and thus allows the system (the bike) to sense some outside stimulus.

Please feel free to define sensor in a different manner if you disagree with [1].

[1] https://en.wikipedia.org/wiki/Sensor


I agree you can have completely non-electronic sensors. Case in point: the sensor wand on a Flying Moth sailboat:

http://www.sailmagazine.com/racing/regattas/learning-to-fly-...


Or, you know, eyes.


This is a soft, squishy octopus robot with a translucent skin that is super-simple and does almost nothing, but as I was watching it on video, all I could think was that, sooner or later, we will have soft and squishy (and perhaps translucent) robots around us, doing all sorts of things. All of this robot's components are soft.

I don't know if it will take 10 or 50 years, but the soft robots are coming; and they will have supple skin and squishy synthetic organs, not metal surfaces and 'boxy' components.


Soft robotics is a bit overhyped. So the two big things that make soft robots interesting right now are the ability not to hurt people if the controller messes up and the ability to squeeze through cracks. If controllers become sufficiently reliable then we lose the first justification. The second is mostly of interest to the military.

There are very good reason to have rigid robots. With higher stiffness one can obtain higher precision and operating speed. The resonant frequency of a robot is proportional to the stiffness of said robot. Higher resonant frequencies imply high operating frequencies and more precision.

Not to mention, most animals, especially land animals, are not entirely soft.


And they will take over all jobs requiring soft skills.


Pleasurebots?


Will still require some hard parts.


This is "autonomous" at the windup toy level.

Festo, the German robotics company, is way ahead in autonomous soft robots. See their "Air Ray" fly around[1] Each year, Festo makes motion controllers and robots, especially high-precision pneumatic devices. Each year, they build something awesome. They do have some rigid components, but if they wanted to build one with fluidic logic, they certainly could.

Their Smart Bird is spectacular.[2]

Festo is a big, family-owned engineering company. 18,700 employees. founded in 1925. Annual sales €2.64 billion. One of the most innovative companies in mechanical engineering, yet a totally different business model than Silicon Valley's.

[1] https://www.youtube.com/watch?v=c3-wIICjAhE [2] https://www.youtube.com/watch?v=nnR8fDW3Ilo


Fluidics [1] has a long history. For better or worse, it has had a very limited use so far. With the advent of 3-D printing, I am hoping to see more interesting projects based on it. On the other hand, fluidics may not be able to stand on its own as a viable technology; it is possible that the most promising ideas will come from combining fluidics, micro-mechanics, electromagnetics, electronics, and optics.

[1] https://en.wikipedia.org/wiki/Fluidics


If that's "completely autonomous", then steam engines are artificially intelligent.

Can it actually walk? The video just shows it waving its arms.


"Autonomous" in the sense that it does not requires external power supply, rather than intelligent.

I agree that the title is a bit missleading, but the rest of the article is clear about this: (e.g. when they say: “Fuel sources for soft robots have always relied on some type of rigid components,” )

Edit: clarity


Nice! but I'm not sure whether we can call that a robot :)


What I don't understand is how this got published in Nature. This work, while cool isn't that novel. Microfluidic logic and soft actuators have been around for some time. Combining the two is not that impressive and more worthy of an engineering journal than nature.


It is my first time hearing the term soft robot and had a hard time realising what a "soft" robot means as "robot" is hardwired to mechanical and electronic parts in my mind. After reading the article, absolutely eye opening!


There's a subfield called soft robotics, which Baymax from the film Big Hero 6 was inspired by: http://www.cmu.edu/news/stories/archives/2014/october/octobe....


I see a lot of people arguing about the significance of soft robotics. What none of the seem to understand is the real significance for most people: The ability to build Baymax IRL.

Seriously, it would be adorable and useful. Somebody help out CMU so that this happens sooner.


Completely autonomous and no electronics?

That's not a robot, that's a child.


Is that considered to be a robot?


Can we change the URL to http://www.eurekalert.org/pub_releases/2016-08/hjap-tfa08161... ? The Eureka Alert page is much more detailed.


We've updated the link from http://sciencenewsjournal.com/new-soft-robot-completely-auto... to the Harvard release page. There was also a previous submission which got a good number of upvotes, but not much discussion: https://news.ycombinator.com/item?id=12341843.


Yeah! I couldn't even find a citation or hyperlink to the relevant article on the "sciencenewsjournal" page.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: