Mobile Microrobots

What about the free-swimming machine? The purpose is no doubt for entertainment. It's entertaining because you have controlit's like a new game. Nobody figured when they first designed computers that there would be video games. So I have the imagination to realize what the game here is: You get this little machine you can control from the outside, and it has a sword. The machine gets in the water with a paramecium, and you try to stab it.

How are we going to make this game? The first problem is energy supply. Another one is controlling the device. And if you wanted to find out how the paramecium looks to the device, you might want to get some information out.

The energy supply is, I think, fairly easy. At first it looks very difficult because the device is free-swimming, but there are many ways to put energy into the device through electrical induction. You could use either electrical or magnetic fields that vary slowly, generating EMFs inside.

Another way, of course, is to use chemicals from the environment. This method would use a kind of battery, but not as small as the device. The whole environment would be usedthe liquid surrounding the device would be the source of a chemical reaction by which you could generate power. Or you could use electromagnetic radiation. With this method you would shine the light on the device to send the signal, or use lower frequencies that go through waterwell, not much goes through water but light.

The same methods can be used for control. Once you have a way to get energy inby electrical induction, for exampleit's very easy to put digits or bits on the energy signal to control what the machine is going to do. And the same idea could be used to send signals out. I shouldn't be telling people at JPL how to communicate with things that are difficult to get at or are far awaythis is far away because it's so small. You'll figure out a way to send the signals out and get them back againand enhance the pictures at the end.

It's very curious that what looks obvious is impossible. That is, how are you going to propel yourself through the liquid? Well, you all know how to do thatyou have a tail that swishes. But it turns out that if this is a tiny machine a few microns long, the size of a paramecium, then the liquid, in proportion, is enormously viscous. It's like living in a thick honey. And you can try swimming in thick honey, but you have to learn a new technique. It turns out that the only way you can swim in thick honey is to have a kind of an "S" shaped fin. Twisting the shape pushes it forward. It has to be like a piece of a screw, so that as you turn it, it unscrews out of the thick liquid, so to speak. Now, how do we drive the screw?

You always think that there aren't any wheels in biology, and you say, "Why not?" Then you realize that a wheel is a separate part that moves. It's hard to lubricate, it's hard to get new blood in there, and so forth. So we have our parts all connected togetherno loose pieces. Bacteria, however, have flagella with corkscrew twists and have cilia that also go around in a type of corkscrew turn. As a matter of fact, the flagellum is the one place in biology where we really do have a movable, separable part. At the end of the flagellum on the back is a kind of a disc, a surface with proteins and enzymes. What happens is a complicated enzyme reaction in which ATP, the energy source, comes up and combines, producing a rotational distortion [here, Feynman is using his hands to simulate a molecule changing shape and experiencing a net rotation]; when the ATP releases, the rotation stays, and then another ATP comes, and so forth. It just goes around like a ratchet. And it's connected through a tube to the spiral flagellum that's on the outside.

Twenty years ago when I gave my talk, my friend Al Hibbs, who introduced me today, suggested a use of small devices in medicine. Suppose we could make free-swimming little gadgets like this. You might say, "Oh, that's the size of cellsgreat. If you've got trouble with your liver, you just put new liver cells in." But twenty years ago, I was talking about somewhat bigger machines. And he said, "Well, swallow the surgeon." The machine is a surgeonit has tools and controls in it. It goes over to the place where you've got plaque in your blood vessel and it hacks away the plaque.

So we have the idea of making small devices that would go into the biological system in order to control what to cut and to get into places that we can't ordinarily reach. Actually, this idea isn't so bad, and if we back off from the craziness of making such tiny things, and ask about a device that is more practical today, I think it is worth considering having autonomous machinesthat is, machines that are sort of robots. I would tether the machines with thin wiresswallowing wires isn't much. It's a little bit discouraging to think of swallowing those long tubes with the optics fibers and everything else that would have to go down so the guy can watch the inside of your duodenum. But with just the little wires, you could make the device go everywhere, and you could still control it.

Even the wires are really unnecessary, because you could control the machine from the outside by changing magnetic fields or electric induction. And then we don't have to make the motors, engines, or devices so very tiny as I'm talking about, but a reasonable size. Now it's not as crazily small as I would likea centimeter or one half of a centimeterdepending on what you want to do the first few times, the scale will get smaller as we go along, but it'll start that way. It doesn't seem impossible to me that you could watch the machine with X-rays or NMR and steer it until it gets where you want. Then you send a signal to start cutting. You watch it and control it from the outside, but you don't have to have all these pipes, and you aren't so limited as to where you can get this machine to go. It goes around corners and backs up.

I think that Hibbs's "swallowable surgeon" is not such a bad idea, but it isn't quite appropriate to the tiny machines, the "infinitesimal machines." It's something that should be appropriate for small machines on the way to the infinitesimal machines.