Claire Corlett

Fish Food, Fish Tanks, and More
Autonomous, self-contained soft robotic fish at MIT

Autonomous, self-contained soft robotic fish at MIT


We are developing soft-bodied robots and there are three important things to know about these robots. First, their bodies are made out of soft silicon and they can bend and twist because of that. They’re also inherently safe to be around. Second, because of their bodies capability to bend and twist, these robots are capable of very compliant motion and they’re also capable of very rapid agile maneuvers which pushes the envelope on what machines can do today. And, thirdly, the robots are self-contained and autonomous; in other words we can package the power source, the computation and the actuation and sensing needed for these robots to deliver their motions. Traditionally soft robots have been either self-contained or capable of high-performance, but not both. So specifically in our lab we want to achieve both of those goals simultaneously in one machine. Currently a soft robot has two parts. One, which is a little bit smaller, is the rigid part where we store all the supporting hardware. And the second part, which is a little bit larger, is the soft body where all the continuous, natural movement happens. And so when we thought about it, a fish made sense. It has a very similar structure: in the head of the fish where the brains are held, it is a little bit more rigid, but in the rear of the fish where the angulatory motion happens, it’s quite soft and compliant. This is our soft robot fish. Like we said, he has a soft body here in green and the supporting hardware up front. The way this fish works is it stores fluid onboard, in the form of a gas, and then releases this gas through a series of pipes and valves into the body. If you think about it, it is very similar to blowing up a balloon. In that case, your mouth would be the pressure source and the balloon would be the body actuator. And basically by inflating and un-inflating different parts of of the body we can get it to angulate. What is special about this fish is it has its brains onboard too. So if I, from my computer, tell the fish to move forward a signal is sent wirelessly through the water to the brains and then the brains tell the hardware what to do in order to move forward. Biological fish use the ‘escape maneuver’ or the ‘c-turn’ to escape prey and they do these maneuvers very fast; on the order of 100 milliseconds. Our robot fish is also able to execute this escape maneuver at the same speed: 100 milliseconds. The fact that our fish can perform an escape maneuver is really important for the field of soft robotics. It shows that soft robots can be both self-contained and capable of high performance. The maneuver is so fast and it has got such high body curvature that it shows soft robots might be more capable than hard robots in some tasks.

100 comments on “Autonomous, self-contained soft robotic fish at MIT

  1. What is the purpose of making this fish robot? If you would use it on open sea wouldn't it be dangerous if a larger fish eat it? 

  2. Uh, it didn't turn really fast, it just bent really fast.  Also a compressed air powered robot seems cool, but it seems a poor choice for an underwater device.

  3. If you really want to learn something, then go to an institute of technology. NOT a theory-based university like Harvard or some other crap. There's a reason MIT ranks ON TOP of theory-based Ivy-league "schools".

  4. Yeah – am I not allowed to enjoy movies, or a sense of imagination at age 21? Sounds like I'm more human than you.

  5. yay, birds, jellyfish, horses, mice, octopus, spiders, dogs, now fish? what robot imitation will we come up with next?

  6. i saw an Xbee in there. Wouldn't that mean the fish is in fact not autonomous but remotely controlled via wifi rather than making its own decisions based on sensor inputs or gps coordinates etc. ? 
    Either way its a great idea for a soft robot ! How far do you guys figure could it travel with a charge of gas? (it looked like it put out quite a bit of bubbles)
    How does it handle unpredicted variables ?
    What sensors did you integrate ?

  7. So why can't these brilliant students and graduates from the best engineering schools in the world (MIT, Cal Tech, Stanford, Oxford) design automobiles that gets at least 200 miles per gallon fuel efficiency?

  8. Hey Professor,
    If you are so rigid and do not blink, you look like a robot.
    Hey Andrew,
    Please, do not put your fist in front of your mouth like that.

    Other than that, nice work.

  9. What happens when this lil device is paddling around, doing robot stuff, then BAAAMM, a fuckin shark comes along and eats him 🙁

  10. They say it uses on board gas to to inflate the soft portion of the robot, but why not use the liquid that the robot is in. as long as the instruments have power it could have a limitless amount of fuel.

  11. Great, but your using the fishing wire to keep it from sinking. How do we also not know if your in fact also pulling the fish forward with it? Put it in an actual environment with no outside influence and see if it actually works next time.

  12. OH that will be a fine addition to the American ROBOT ZOO we seem to be un willingly paying for.
    right along side The Human hunting 60mph DARPA CHEETAH.

    now tell me  again why I need these things …..

  13. Hello MIT, just happened to see your video and recalled all the crap you caught for installing a car elevator in your house; can't imagine what they would have said if they knew you were working on a robotic fish! Take care.

  14. It is a good beginning but you need to lose the bubble machine to make it turn. Fish do that with their tails and fins, not blowing bubbles out their gills. Completely blows your I'm pretending to be a fish thing.

  15. 1. Cost? Is it patented?
    2. Is there a camera(s)/gps already fitted? If not, how easy to add one or more?
    3. Total weight?
    4. Limitations to built a shark sized model?

  16. Just imagine letting one of these go out into the ocean, there can be two outcomse:

    Letting an expensive piece of technology get chomped up by a predator such as a shark or whale

    or

    Mating with other of its species to produce cyberfishes and plotting to take over the fish race and possibly the human too.

  17. Why let the air escape?   Really it could be in a circuit, or send a puff of air one way and then let it back to flex  and unflex one side.   Still it is impressive for an early effort.   The main thing in science that I have learned over the years is that most things are reversible.  If compressed air can make the "fish" flex,  a  "flag" flexing in the wind with pipes in much the same place inside the flag,  can compress air!  (I have already dreamed up sails going too and fro in unison, like a shoal of fish, on clothes lines, compressing air to do  useful work).  

  18. Oh no thanks Obama, keep that science fish out my river! I dun want to be fishing one day, catch one of these, clean it, take it home, cook it, butter it up and take a bite ONLY to see its a robot fish I broke my tooth on?!? 

  19. Who cares about an Underwater Green Gas Filled Soft Robot Fish???

    Good God!!!!

    Wake Up Lab Rats 

    Where are our Flying Cars and our Robot Butlers?

    Robotic soft fish???

    If I want my toilet cleaned i'll tell my robot butler to do it.

    Can't get a toilet smelling robot fish to make sandwiches now can we?

  20. I'm sorry but that robo-fish has very clumsy movements, not even close to the real thing, maybe it can be use like as a lure, but even the real fishes will not get fooled with that rubber retarded thing.

  21. Scientists are trying to do ugly clumsy robot fishes, while Jesus christ put hundreds of fishes inside the fisherman's net, HA take that, Science 0, Religion 10+

  22. It needs to be suspended by a wire to prevent it from sinking.  You would think that if they an design a robotic fish, they could also come up with a buoyancy type mechanic as well….it would seem a lot cooler if they didn't have to hold it up with the wire

  23. its the end of the fucking world people! this shit is out of hand! CIA robot fish will be looking at your girl friends vagina's in a matter of weeks!

  24. There are so many negative comments because this prototype is way too early for demonstration.  You got the mecanism to do the tail motion but that's about it, no self level and dept control.   That tiny C02 tank will not last very long either.  
    Good work.  Still in a very early state of development.

  25. Very cool! But I'm confused…how is a "self-contained" robot autonomous if you have to send it commands from an external computer? For example, in a senior mechatronics course 10 years ago we used the MIT Handyboard to take input from 5 types of on-board sensors to control 2 drive motors and a "stick" motor on a game of 2-on-2 robot hockey. We had a program for offense and defense. The rules stated you hit "GO" and hands off! Isn't that true autonomy? No external devices allowed…

  26. I believe they mean that the robot can move and respond to stimuli on its own if its programming is tweaked. They probably just controlled it with the computer for the video.

  27. Now make it so they run off fish food and the oxygen in water, give them AIs, breed the AIs that last longest, and keep going until you have real life robotic fish. I know that's what I'd do.

  28. If you're going to try to emulate nature, you should emulate nature. For example the squid sucks in water and then squeezes it out for propulsion. Water being all around it, is a lot easier to come by then a refillable, pressurized air tank.

  29. Professor Daniela Rus. I admire you! I was reading many articles of Soft Robots and some of them are yours. Nice job! I want to be a professor too and lead a team in Robotics

Leave a Reply

Your email address will not be published. Required fields are marked *