December 3, 2014 • 37 mins
What if you could own a big robot made up of thousands of smaller robots, and it was able to change its shape in order to complete various tasks?
Learn more about your ad-choices at https://www.iheartpodcastnetwork.com
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Brought to you by Toyota Let's Go Places. Welcome to
Forward Thinking, Either in Love and a forrard Thinking, the
podcast that looks at the future and says, they say,
we are what we are, but we don't have to be.
(00:20):
I'm Jonathan Strickland, I'm Lauren Folk, and I'm Joe McCormick.
So we've got a fun topic lined up today. Yeah,
So it's it's one that I really was thinking about
recently because I saw this awesome movie, Big Hero six.
I haven't seen it yet, so really good. The trailer
was really cute. I haven't seen it either, but Jonathan
keeps talking about it, so I think we need to
(00:42):
go see it. So before I get into what today's
topic is, specifically, the thing I wanted to say about
Big Hero six is the movie opens. It's set in
me an alternate future really, because you have the main
city is San Francikio, so it's a kind of a
melding of San Francisco in Tokyo and are just world design.
But the thing that really impressed me early on is
(01:03):
that there's all this kind of science fiction ee sort
of stuff going on, but a lot of it, at
least has its basis in science. Fact, a lot of
it is is based on you know, take take sort
of the cutting edge stuff that's going on in labs
right now, projected out several years, and assume that the
research you know plays out and that everything is fruitful,
(01:25):
and what do you get? And a lot of the
technology you see, at least in the early part of
the movie seems to go along with that. And I
was just really impressed. I mean, everything from soft robotics
to what we're gonna talk about today, which is self
reconfiguring modular robots or as you also noted that they
can be called in your script for this video episode,
(01:46):
autonomous kinematic machines of variable morphology rolls off the tongue.
So what on earth is that talking about? Well, I
think we should start with the idea of the variable morphology. Okay,
So something that changes shape, right more fall as he
refers to like structure or form or shape as it
would in biology. So so what we're talking about is
(02:09):
a transformer, that's right, Yeah, it's a shape shifting robot.
And why is a transformer? Great, well, a transformer is
it's a robot in the SUI, A transformer is great
unless you're talking about one of the recent movies, a
cartooning transformer or a real transformer. Either one of those
(02:29):
would be great. But it's great because it can do
two jobs, whereas most robots can only do one. So
a regular robot might be an autonomous car and it
can drive down the highway distributing palettes of Captain Crunch.
Excuse me, Captain Crunch the hyphens. I would really hope
you'd be an admiral by now. But then, of course,
(02:51):
after that autonomous truck delivers them, it can't get into
big wrestling matches with other robots. Now, you might make
a robot that can get into wrestling matches with other
robots and stabbed them through the side of a highway
with a giant sword. But that robot can't be a
truck unless it's a transformer, right where it can be
(03:12):
both you can. But but wait, let me do you
one better. What's better than a transformer that can assume
two morphs? A robot that can assume even more morphs?
What about a robot that can assume an infinite number
of morphs? So you're saying, like a robot that is
capable of assuming any shape that you need at any
(03:32):
given time to perform any given task. Bingo. Well that
is something that is is in Big Hero six, the
the hero whose name is Hero actually designs self self
self reconfiguring modular robots. He designs a tiny robotic unit that,
when it connects to others of its kind, make larger shapes.
(03:56):
And you might think, well, that's pure science fiction, but
it's not. And we want to kind of go into
more detail about why such a thing would be a
desirable object in the first place. And you you really
hit on it, Joe, the idea that robots today tend
to be really good at specific things and crappy at
everything else. Yeah, they have one job. You had one, Joba,
(04:20):
I asked you to paint a masterful painting. You had
one job, and all you did was clean the floor. Yeah,
that's exactly right. You know, we do. We design robots
to perform specific tasks, and so the form follows the function. Yeah.
In fact, there almost seems to be a direct proportional
correlation where the more diverse a robots skill set gets,
(04:42):
the worse it gets at all of those jobs. Instead
of can be really good at one or very mediocre
at several times. This is like that TV trip about
about ninja and that if you have one ninja, it's great,
and if you have a hundred ninja, they have about
the same capacity. Is that one ninja would have just
spread out amongst all of them, right right, That's why
(05:03):
the tick can just breathe right through a horde of ninja. Yeah. Yeah,
besides the fact that he's also ni invulnerable. Um yeah,
they the this is that's exactly right. Yeah. That you
build a robot that you want to be able to
do lots of things. Usually it can do a few
things and not so great. And this is you know,
anyone who's who's played with any of the robots that
(05:26):
are social bots that are meant to do like these things,
like mostly in the toy department, you know, they're amusing,
but they're not exactly the best at anything that they're
trying to do, Okay, And there's solid design reasons for
creating robots that are that that only have one job, right,
I mean, creating robots is kind of expensive and time consuming,
(05:47):
and so if you're trying to build lots of different
functionality into them, that's that's a lot more expensive and
a lot more time consuming if you if you're narrowing
down what you wanted to do, right, if you if
you pinpoint that focus, then you say, here's what the
robot has to be able to do. Anything that does
not directly contribute to this task, we can eliminate. That
(06:09):
means that you can, you know, really pinpoint that focus,
really make sure that that design is as efficient as
possible at doing that task. But let's imagine a scenario. Okay,
let's say that you live in a house by yourself,
and you work long hours, and you need a robot
to help you clean the house. Beyond beyond something like
(06:33):
a roomba that's just going to sweep the floor. Sure,
a roomba might be really good at sucking up dust
off the floor, that's its one job. But a roomba,
of course, cannot clean your gutters, and it can't well
I don't know, I was gonna say, it can't soap
and mop the floor. But they might make rumas that
can do that. They do. They do make mopping room
bas but they're not usually also sweeping room bas So
(06:54):
it's in other words, you have you have two different
robots that have similar but different jobs. Okay, and one
about another one thing. Yeah, another one for washing windows,
and another one for doing your dishes. And there there
have been lots of robots that do these things, but
again they're all individual like their uni taskers, as Alton
Brown would call them. So are you going to fill
your house up with thirty five different kinds of robots?
(07:17):
Well maybe if you can afford it and you don't
mind occasionally getting sharp metal jabbed into you. Well, also,
you just you started this up by saying you live alone.
I mean that maybe they'll make you a little less
lonely to have like thirty five robot friends doing all
your work for you. That would make me more lonely,
would it? Yeah? I just named them. I assume if
you live some with somebody else, you can bully them
into doing all the houses. Is that how it works
(07:39):
in your household? Kidding? My wife is wonderful. Yeah, I
do the same thing, Joe. Uh yeah, so this is
this is exactly the point you're making, though, is is
very much on target right. The idea that it would
be very beneficial to have a mod geiller robot, a
(08:01):
robot made up of smaller robots, smaller pieces that could
form any given shape to perform any task as best
as it possibly could, rather than have a veritable army
of robots. I mean, even if you were to say, well,
let's imagine that I've gotten enough space for all those
robots and I would want them all. Just the power
(08:21):
bill alone to keep them all charged would be ridiculous.
So what if you could have this group of tiny
robots that joined together to form a much larger moving
object that can do pretty much anything you wanted to do? Right, So,
it could form into a small disc and hoover along
the floor like the room but does. Or maybe it
(08:42):
could make itself tall, rearrange itself into a long stick
with arms to clean the windows, or to reach into
the sink to clean the dishes. Or it could even
climb over upon itself up the wall, up the side
of your house to reach the roof and clean out
your gutters. Yeah, yeah, all of these are I mean,
it sounds kind of science fiction e but these are
(09:02):
all real things that, uh that engineers are looking into.
Maybe not necessarily those use cases, but but essentially the
idea of a robot that can change its shape in
order to do different things. Whether that's just to get
across a landscape or to actually perform a specific task. Right, Well,
I think we should talk about some of the existing
(09:23):
models of reconfiguring or self reconfiguring modular robots that exist
out there today, right because obviously we don't have anything
like what we just described, this perfect self reconfiguring robot
in your house, but we we we've basically got a
few steps in that direction that are very encouraging. Yeah,
we've got to that we're specifically going to talk about,
(09:45):
one of which is a reconfiguring robot, one which doesn't
necessarily reconfigure, like it doesn't form a larger robot, but
it can act with a swarm intelligence, which it will
be important in modular robotics. So we're gonna talk about those.
But I should point out there are lots of different
types of modular robots out there. Some of them are
not self reconfiguring. Some of them are simply you get
(10:07):
all the different components and you put them together. Yeah
you can. You can sit there and make you know.
There are a lot of ones that teach children the
basics of robotics, and they have different units that do
different things, and the way you put them together determines
their function. Uh, these are modular, but they don't self reconfigure.
They don't they don't manage to put themselves into whatever shape.
(10:29):
You are the one responsible for doing it. So we're
not really going to talk about those. Okay, Well let's
look at M I T S M blocks. Yes, they
are momentum driven magnetic modular robots. That's how the commercial
would go. You know, I imagine the guy who did
the micro machines ads doing it. Uh no, no, I know.
How the commercial would go would be like, let M
(10:51):
into your life. The M block, you know, M I
T you can just send the checks are away. So
they're cubes that they you would think because they're called blocks,
Joe's already regretting absolutely going to stay in. Okay, So
(11:12):
the M blocks are these blocks that have magnetic corners, right,
so each corner has a beveled magnet in it, which
is pretty cool. I'll talk a little bit more about
that in a second. They also contain a rotating flywheel
that can move up to twenty thousand revolutions per minute.
It's pretty fast, and so, uh, they don't necessarily speed
(11:34):
turn at full speed every time, but depending on what
you need the cube to do you might have it
speeding spinning that quickly, and then what you do is
you apply a break. Now, the laws of momentum state
that that energy has to go somewhere. It gets transferred
to the cube, and this makes the cube move. Okay,
so what does this actually look like when we're watching
these cubes move around on each other? Have you ever
(11:56):
seen a ghost movie where objects seem to be moving
on their own accord with nothing obviously moving them forward. Yeah,
that's kind of what it looks like. It looks like
the blocks are possessed, kind of like kind of like
the toaster jumping and Ghostbusters too. Yeah, yeah, exactly. So
I don't know if you already said this or not,
but they don't have external moving parts. No, it's all
(12:18):
it's all kept inside. The flywheel is actually contained within
the cube, right, So the momentum can move one cube
onto another cube and reposition it around the outside of
that cube without attaching and reattaching via clamps or hooks
or anything like that. But it's it's attachment through those magnets, right. Yeah,
And so if you imagine one it's just one cube
(12:39):
moving around another. But if you add a third, you
could have two cubes moving around on each other, and
then the fourth or fifth, and eventually you can end
up with a large mass of moving modules that can
assume shapes, especially if they're in communication with each other.
And you can also include modules that don't necessarily have
the flywheel inside them. They might have some other UH
(13:02):
specific sensor or camera in them. Uh. The m I
T Group has talked about the possibility of that as well,
in which case the ones that have the flywheel would
be responsible for moving the other ones, which they can
do because that flywheel spinning so quickly. Um, these things
can actually jump, so they don't just like the the
videos I've watched show the cubes. I guess rolling is
(13:23):
the best way of putting it in order to move
across the floor, but they can even Let's say you've
got two cubes stacked already vertically and you have a
third cube roll up. It can actually spin and break
in such a way to jump up and land on
top of those two cubes, which is kind of spooky
looking when you watch it the first time. They also,
(13:43):
of course are making pretty loud noises because it's a
flywheel spinning, and then breaking, so it's it's not silent,
it's it's loud and spooky, so but it's pretty cool.
And and those those magnets that are in those beveled
points in the corners, they rotate. And the reason by
the magnets rotate is so that you can always get
a north south connection. Remember, you know, the opposite poles
(14:05):
attract and similar poles repel, and you want to be
able to make sure that when one block approaches another block,
it's able to actually latch on properly. So the magnets
in those individual points will naturally rotate so it will
be a north south connection so that any two cubes
from any two sides can attach exactly. And if you
get two poles that are the same, then one of
(14:27):
them will end up rotating. Everything will be fine. So
there's literally no wrong way to put them together. Neat. Yeah. Yeah.
These also remind me of another project out of m
I T when we've talked a little bit about before, previously,
before previously, which is the common way to really over
emphasize the point um. They're the m I T S
(14:48):
reconfiguring robots from their self assembly lab um while researching
what's called forty printing. Um. That's three D printing of
materials that can then move and change with with minimal
stimul lists or instruction from the outside. Uh. The kids
of the Self Assembly Lab created some macro scale robots
to help demonstrate the usefulness and honestly the coolness of
(15:09):
this changeability of of structure and so so these are
these reconfiguring robots are programmable folding chains constructed of of nodes,
each of which is hardwired to one or more neighboring nodes.
So if you give one note an instruction, it will
pass information on how to move down through the chain
(15:30):
until the programmers desired shape is is complete with with
relatively minimal input from from that programmer. Right. If you
do not remember us talking about it, or or if
you want to refresh yourself, you can learn more about
these and other self assembly techniques in our video and podcast,
both of which published August. That's like over a year ago. Guys,
(15:52):
we've been doing this a minute. Um. The episodes are
called the video is forty Printing is the Future of Design,
and the pod cast episode is forty Printing is One
D Better. I must have titled that one, actually, I
think I came up with that line. Strangely enough, Yeah,
very it's a very Jonathan kind of joke. So the
(16:15):
other one we wanted to talk about was Harvard's kill abots. Now,
this is the one that is not modular robot but
they do demonstrate swarm behavior, which again is going to
be really important if you're talking about ultimately, if your
if your goal is to create a robot, a large
robot made up of smaller robots, then you have to
build in some form of intelligence for each of those
(16:36):
smaller robots to know quote unquote its place and it's
uh and it's function right. And so there are some
similar concepts going on between these two uh, these two
ideas swarm robotics and the modular robotics. So with a
self reconfiguring modular robout, we're generally imagining something that's connected
to itself. It's got pieces that may all be the
(16:58):
same kind of peace moving on one another to reshape itself.
With swarm robotics, you've still got lots of pieces working together,
but they're sort of moving on mass and they might
not be building a shape, right, they might they often
aren't physically connected together at all, right, they're all individually
free to move. If if somehow one was to break
(17:20):
free of its programming, it can make a break for it. Uh,
it's not. It's not chained physically in any way. So
whereas the ones from M I T were magnetic and
linked together, and we know of a lot of others
that have physical connections of some sort, either they have
connectors that snap in or they have clamps that hold
onto one another, these do not. These are simply to
(17:42):
allow UH computer scientists to test various algorithms for robotic intelligence,
for swarm intelligence. UH and a lot of them are
looking into bio mimicry, the idea of of how animals
and swarms behave in a mass. You know, it's not
it's it's really fascinating stud At any rate. The problem
(18:02):
with these computer algorithms was that for a very long time,
there wasn't an easy way to test them. Like you
might come up with what you think is a brilliant
algorithm to to inform a swarm of robots how to
move for any given purpose, and you can test that
through computer simulation to your heart's content, which is not ideal, right,
(18:23):
I mean, because you're depending upon the fact that the
computer simulation has to be accurate enough to actually reflect
what a physical object would do. So Harvard scientists came
up with this idea and professors as well came up
with this idea to create a relatively inexpensive tiny robot
on which you could test these algorithms by by assembling
(18:44):
enough of them to do whatever it is you need
to do. And they look like, well, they're little circles
like think of about the size of a quarter. But
they are on long, spiny legs and there's a little motor,
a vibrating motor on them that depending on how it's
vibrates the assuming that this is on a level, smooth surface,
(19:05):
the robots will skitter across the ground. Now, their their
legs aren't moving, right, it's just through this vibration, so
it's not like they're segmented legs. It's not that much
like an insect. But they do jitter so that they
move in a specific direction. Yeah, well, these things certainly
do that, And there's some great videos online of these
(19:27):
moving around in various ways. Usually it's a direct overhead shot,
so you just see a little it looks like a
hockey puck, just kind of slowly moving across an area,
so you can even make these yourself if you really
wanted to. I'll tell you more about that in a second.
But the way the way the kilbots form that shape
is all dependent upon the algorithm. Assuming that the algorithm
(19:48):
was was properly constructed so that the robots know what
to do quote unquote um, that they are able to
detect one another and move according to whatever the plan was,
and they communicate through various means like there's uh. The
version I saw, they were specifically using infrared emitters and
sensors so that they would be able to tell how
(20:09):
far apart they were from their neighbors to make sure
they maintained whatever the distance was that the algorithm dictated.
So one of the examples I saw was a follow
the leader um example, where one was designated the leader.
They had actually put a green sticker on top of
it so it made it easy to follow on the video,
and then the next kill abot had the instruction to
(20:31):
try and maintain the same distance with the first one.
The first one's instructions were move wherever you want, but
you can't get too far away from the second one.
The third one's instructions were tried to keep up with
the second one, but don't get too far away from
the fourth one, etcetera, etcetera. So the longer the chain is,
each one essentially has the same instruction maintain as as
(20:53):
you know, as as a stable a distance as you
can to the one in front of you without getting
too far away from the one behind you, and it
would follow the leader this way. And it was a
cute little demonstration. They also had a foraging demonstration where
they had one kilobot designed to look like uh, well,
it just had a red sticker on it. Those designed
to act as the nest this is home base. And
(21:15):
they had another kill abot that was pound off maybe
to three ft away, that had a green sticker that
represented food. So if they wanted to have these robots
go out and search for So we're teaching robots to
be cannibals. Well, it's much better than teaching them to
hate humans. I'm not worried about the killers. Things skitter
(21:38):
at me all they want. Yeah, they're only called killer bots.
And how one letter off from kill bots? Yeah, so
uh yeah, and and well, at any rate, so the
way this would work is that the once they got
the instruction, they used their infrared sensors to look for
the quote unquote food kill and once they found it,
(21:59):
then they would travel back and forth between the food
one and the next one. Uh. Some would just end
up forming kind of a structure around the nest. Uh.
Once they were unable to find the food, they just
kind of stayed still. They wouldn't continuously move around, they
would only go so far away. And that was a
kind of cool demonstration as well. Now, like I said earlier,
(22:21):
you can actually get the plans on how to build
one of these yourself, or lots of them, since one
would not be terribly useful. You can't really test swarm
intelligence on a single robot. The it's difficulty. Yeah, all right,
just pretend you're a lot of people. Uh. But the
whole design is published under Creative Commons. It's just a
non commercial commercial license, so you can't you know, you
(22:42):
obviously can't make them to sell them, but for educational
purposes you can make as many as you can afford
to build. And I say afford to build, because obviously
you have to buy all the hardware to put these
things together. Though, you know, if you're building them in bulk.
Obviously the cost will go down. Yeah, although it's still
pretty expensive because if you were to build say about
a hundred, they price it out as being about fifty
(23:04):
dollars per robot. It's pretty expensive. If you were to
build a thousand, it's twenty dollars per robot, But that's
twenty tho bucks when you take that all into consideration,
and I don't have twenty tho dollars to build my
robot army yet, so uh but at any rate, again,
these kind of test out that idea of swarm intelligence.
More recent videos have shown them do things like assume
(23:25):
specific shapes, like they give it a command to for
the all the robots and I think there were in
this case more than a thousand of them in the
demonstration I saw, but to form a shape like a star.
And so the robots have to have information about which
other robots they're supposed to be next to and how
(23:45):
they're supposed to be configured based on that, which is
really fascinating. You think, you know, this thing is again
about the size of a quarter um. I mean, obviously
it's thicker than a quarter is, but still they're able
to pack in enough of the microprocessor technology for the
robots to be able to determine, oh when when they
we make this shape, I'm number fourty seven. I have
(24:06):
to make sure I'm next to four and three, six
and four and thirty eight. And they do kind of
go one at a time to make this shape you
like it was it was like in a line, as
opposed to all right and break didn't They didn't do
it like that. But it was really interesting stuff. And again,
this is the kind of behavior we're going to have
to see for something like the modular robot that's in
(24:27):
Big Hero six to become a reality. Well, I think
this stuff is awesome, and I want to know what's
currently in development. I don't know if you've looked into that, Jonathan.
I looked into sort of the most recent literature, because
this is one of those things where we still have
people working on implementations. No one, I don't think anyone
has come up with the ideal implementation yet. It's it's
(24:50):
of course not I mean, we're in the we're in
the research phase. Yeah, exactly. This This is a developmental era,
which kind of makes it even more exciting really, because
you know, there's there's so many people looking at different
ways of doing this in different tasks that such robots
could do. That it really starts to fire your imagination.
So for example, here's here just some titles of some
(25:12):
recent scholarly articles that are on the subject. Uh and
and they kind of demonstrate now and like by recent,
I mean these were all published in two thousand fourteen. Uh,
these kind of demonstrate the the breadth of the research
design of modular robot system for maintenance tasks and hazardous
facilities and environments. Well that that kind of you know,
(25:35):
you can easily imagine having a modular self reconfiguring modular
robot would be incredibly useful in that source of situation.
Or how about room bots. A hardware perspective on three
D self reconfiguration and locomotion with homogeneous modular robot. So
here we have a specific scholarly look into well this
is you know, we we know what we want to achieve.
(25:58):
What are the way is to do it? Using this
specific approach where you've got robots that are all essentially
the same shape that combined with one another, then we
have design principles and testing of a latching modular robot
connector So again looking at different ways for these different
modules to join up exactly. H then you know the
(26:19):
modular robot joint design and experienced verification for small openings
and enclosed space, which again really awesome because you're you're
looking at the possibility of using these robots that could
go into places that you know, if it had a
solid form factor that didn't change, it couldn't get there.
So imagine that you have to send uh, this robot
(26:40):
into a building that is you know, perhaps it's a
really old building and it's got U faulty structure, you know,
structural damage, and you want to make sure it can
navigate all the way through and actually investigate that structural damage.
That could be really useful. Sure, yeah, if there's a
collapsed path or something like that and you need to
get under it or round it or whatever that it is,
(27:01):
that that is right right, being able to break down
into all of its creepy little robot parts and skirt
her on through is terrifically useful. You know, it's stasis
on the terror. You know. It strikes me that there
are sort of like two main ways that this technology
can move forward, and really both of them seem crucial,
but one of them is about working on the modules themselves,
(27:24):
like making each individual piece, and we're assuming we're working
with robots that are made of pieces that are all
identical pretty much, Uh, that making each individual piece cost
effective but very versatile, able to move around, sturdy, able
to communicate, without being a gazillion dollars apiece. But then
also talking about the overall swarm intelligence, how do you
(27:48):
program the robot through these modules to have some kind
of collective brain that enables it to do its work effectively? Right? Right? Yeah?
And this is I mean, these are a lot of
unanswered questions right now, right can could there be an
emergent artificial intelligence from the swarm intelligence that would allow
(28:08):
you to essentially tell it here's what I need you
to do, you figure out how to do it. Or
would you always need to have some sort of controlling
system that would essentially send instructions to that swarm of
robots to form a particular shape to complete that test.
Oh yeah, that's an interesting approach. I hadn't even thought
(28:29):
about that because I was just imagining we're assuming this
has to be autonomous in one way or another, but
you could just I guess, have a laptop that's bluetoothed
up to your modular robots swarm, and this thing is
reconfiguring itself to do work based on orders that are
sent to it from the software on your computer. Yeah. Well,
I mean when it comes down to it, the question
(28:51):
is how autonomous does it get? Because there's some autonomy
that is necessary, right, because if you're talking about thousands
of robots, you don't have the time to tell every
single individual robot where it fits in this grand scheme. Sure. Sure, well,
I mean you know, if you think about something like
like having a a lead butt with a green sticker
on it or what have you, um, that you can
bluetooth up with and give a command too, and then
(29:13):
have it break that down into commands to give to
everyone else in the swarms that you don't have to
literally contact every single microbot, right. I think there will
be a bridging system. Yeah, I was assuming the commands
would be given by software, not like manually by well well, sure,
I mean you know, no EI either way though, Um,
there are lots of different factors of trying to communicate
(29:34):
with a swarm like that, and how you would go
about making it the most efficient, and it's kind of
like when we talked about neural networks, how every time
you add a node, it complicates the entire structure. I
don't want to say exponentially, but quite a bit large fold.
So same sort of thing with modular robots. The more
modular robots that are in the system, the more complex
(29:56):
it's going to be to command it. But that's what
you know that swarm and eligence is all about. So
that's really exciting and uh, you know, kind of looking
into the future, like let's assume again kind of like
Big Hero six did. Let's assume that this this research
pans out and that we come up with some implementation.
It doesn't you know which which one is immaterial for
(30:18):
this discussion, but some implementation that works. And then in
the case of Big Hero six, Uh, it was all
magnetics that held everything together, and the individual modules look
like a sphere with two little uh protrusions that stuck
out that could that were a kind of hinge. They
could move independently of each other. So uh, that's what
(30:39):
the individual component was, and then together they could form
pretty much any shape you you wanted. Um, and they
did so using a brain computer interface, which was really cool.
Is essentially a headband you could wear where your thoughts
get transmitted into computer commands that tell the robots what
it needs, what they need to do. Uh, that's awesome. Anyway,
(31:04):
we're quite a ways away from that too, But people
are working on brain computer interfaces just as they're working
on this modular robot issue. So you would want, in
an ideal implementation to have your microbots as small as
you could possibly make them, right, because that that would
increase their versatility. But also, you know, you can think
of it kind of like an image of digital image
(31:26):
and in pixels. You know, if you have smaller pixels,
you have better resolution. The picture looks better as opposed
to large blocks of color than it looks blocky and
and jagged. Same sort of thing with your modular robots.
The smaller the unit, the more smooth the robot's going
to be, the more versatile it's gonna be. Um, but
that also comes with some pretty big issue problems. But
(31:48):
another one is that this ideal implementation would allow it
to move through those small spaces we were talking about,
whether individually, you know, so it splits apart and all
the individual pieces creep through an area, or it forms
some other shape, like a snakelike shape to get through
certain areas. I've actually seen some really awesome snake bots
that are incredibly disturbing to watch on video. Uh, we'll
(32:12):
probably need to power these suckers with something that we
haven't thought up yet. Yeah, this is a real problem.
Oh sure, you know, well, any significant pushing or pulling
or lifting or other interaction requires really a lot of
power right now. Hydraulics are really the most efficient systems
for for that kind of stuff, and they're you know,
for example, what what drive the most badass exo suits
(32:35):
for for lack of a better term, um, but they're
also really clunky and don't really lend themselves to this
microbot kind of format because you need like a tank
of fluid in order to drive. I can't imagine having
like super small lines to each microbot and still having
any kind of degree of freedom. Yeah, it would be problematic.
Yeah yeah, so I mean, you know, maybe the answer
(32:55):
to this is that we were not going to use
this kind of system for any kind of heavy lifting. Um.
But even so, just powering small things that you want
to keep lightweight is a problem. All the time. We
talk about it always add weight. So one other challenge,
and we've talked about this, is the fact that we
need to have these these individual units be able to
(33:19):
communicate with each other at least on some basic level
so that they quote unquote know what they're supposed to
do and where they're supposed to go. And that's where
we're seeing this this work with, like in the Harvard
kill abuts of of Swarm Intelligence. It's it's one thing
to see the work that's being done with things like
the the M blocks where you're seeing maybe six or
(33:41):
seven of the units working together. It's another thing when
you start talking about you know, large collections of these devices.
We talked about that a little bit before too, in
our episodes about ants and also about biommicry as a
whole um and also in our upset about right emergent behavior. Yeah,
(34:02):
these are all things that you know, we don't have
answers for because we're still asking the questions. But it's
really exciting again to see the work and while knowing
about the stuff because you know what, we've researched so
much into robots and artificial intelligence just for this show. Uh,
the not not just this episode, but the forward thinking
series overall that when I went in and saw this movie,
(34:25):
when I was watching Big Hero six, That's why I
was like, I had a huge grin on my face
because I was thinking, these are all the things we
talk about on our show. Uh that where we are
now in a world in this movie where all those
things have been realized, where all the problems and challenges
we talk about have been overcome, people have figured out
the way to engineer a system that meets those challenges,
(34:48):
and to look at the potential of that world was phenomenal.
I was like, Ah, it's like I'm watching the outcome
of a forward thinking podcast where we have that bit
where we say, what will the future like? And this
is what the future would be like? And it's it's
an amazing vision of that future. So um yeah. The
first ten minutes of that movie, I was like, man,
(35:10):
I wish they had called me because this is awesome.
It feels like it's the outcome of a forward thinking show.
Um but you know, and then of course it gets
into more fantastical elements as it goes on, and and
the science and technology get a little more loosey goosey,
but they had already won me over at that point,
it didn't matter. So you have a hard heart for
(35:32):
scientific implause. There's I can get pretty nitpicky. I admit
it wouldn't happen like that. Yeah, yeah, well, I mean
they're there. Definitely comes a point in that movie where
you gonna say, all right, I just gotta go along
with the ride Bernell or I want to be it's
it's it's worth seeing you guys should definitely see it.
And of course, you know the Smudgler robot research. We're
(35:55):
going to keep an eye on it. It's fascinating stuff.
We're learning so much about robot intelligence and robot mobility
through this kind of stuff that it's just a fun
thing to keep our eye on. So, guys, if you
have any suggestions for future episodes, maybe there's something else
about robotics you've always wanted to hear about. Obviously we
don't shy away from that topic. Let us know. Or
(36:17):
if there's anything else you know, just something you're wondering about,
what would that be like in the future, send us
a request. We read all of them, we're happy to
get them, and we would love to hear what you think.
Send us an email addresses FW thinking at how Stuff
Works dot Com, or drop us a line on social media.
We are on Twitter, Google Plus, and Facebook. At Twitter
and Google Plus, we used to handle f W thinking.
(36:38):
Just search fw thinking and Facebook. We'll pop right up.
Leave us a message, and we will talk to you
again really soon. For more on this topic in the
future of technology, visit forward thinking dot com, brought to
(37:02):
you by Toyota Let's Go Places,
Brought to you by Toyota Let's Go Places. Welcome to
Forward Thinking, Either in Love and a forrard Thinking, the
podcast that looks at the future and says, they say,
we are what we are, but we don't have to be.
(00:20):
I'm Jonathan Strickland, I'm Lauren Folk, and I'm Joe McCormick.
So we've got a fun topic lined up today. Yeah,
So it's it's one that I really was thinking about
recently because I saw this awesome movie, Big Hero six.
I haven't seen it yet, so really good. The trailer
was really cute. I haven't seen it either, but Jonathan
keeps talking about it, so I think we need to
(00:42):
go see it. So before I get into what today's
topic is, specifically, the thing I wanted to say about
Big Hero six is the movie opens. It's set in
me an alternate future really, because you have the main
city is San Francikio, so it's a kind of a
melding of San Francisco in Tokyo and are just world design.
But the thing that really impressed me early on is
(01:03):
that there's all this kind of science fiction ee sort
of stuff going on, but a lot of it, at
least has its basis in science. Fact, a lot of
it is is based on you know, take take sort
of the cutting edge stuff that's going on in labs
right now, projected out several years, and assume that the
research you know plays out and that everything is fruitful,
(01:25):
and what do you get? And a lot of the
technology you see, at least in the early part of
the movie seems to go along with that. And I
was just really impressed. I mean, everything from soft robotics
to what we're gonna talk about today, which is self
reconfiguring modular robots or as you also noted that they
can be called in your script for this video episode,
(01:46):
autonomous kinematic machines of variable morphology rolls off the tongue.
So what on earth is that talking about? Well, I
think we should start with the idea of the variable morphology. Okay,
So something that changes shape, right more fall as he
refers to like structure or form or shape as it
would in biology. So so what we're talking about is
(02:09):
a transformer, that's right, Yeah, it's a shape shifting robot.
And why is a transformer? Great, well, a transformer is
it's a robot in the SUI, A transformer is great
unless you're talking about one of the recent movies, a
cartooning transformer or a real transformer. Either one of those
(02:29):
would be great. But it's great because it can do
two jobs, whereas most robots can only do one. So
a regular robot might be an autonomous car and it
can drive down the highway distributing palettes of Captain Crunch.
Excuse me, Captain Crunch the hyphens. I would really hope
you'd be an admiral by now. But then, of course,
(02:51):
after that autonomous truck delivers them, it can't get into
big wrestling matches with other robots. Now, you might make
a robot that can get into wrestling matches with other
robots and stabbed them through the side of a highway
with a giant sword. But that robot can't be a
truck unless it's a transformer, right where it can be
(03:12):
both you can. But but wait, let me do you
one better. What's better than a transformer that can assume
two morphs? A robot that can assume even more morphs?
What about a robot that can assume an infinite number
of morphs? So you're saying, like a robot that is
capable of assuming any shape that you need at any
(03:32):
given time to perform any given task. Bingo. Well that
is something that is is in Big Hero six, the
the hero whose name is Hero actually designs self self
self reconfiguring modular robots. He designs a tiny robotic unit that,
when it connects to others of its kind, make larger shapes.
(03:56):
And you might think, well, that's pure science fiction, but
it's not. And we want to kind of go into
more detail about why such a thing would be a
desirable object in the first place. And you you really
hit on it, Joe, the idea that robots today tend
to be really good at specific things and crappy at
everything else. Yeah, they have one job. You had one, Joba,
(04:20):
I asked you to paint a masterful painting. You had
one job, and all you did was clean the floor. Yeah,
that's exactly right. You know, we do. We design robots
to perform specific tasks, and so the form follows the function. Yeah.
In fact, there almost seems to be a direct proportional
correlation where the more diverse a robots skill set gets,
(04:42):
the worse it gets at all of those jobs. Instead
of can be really good at one or very mediocre
at several times. This is like that TV trip about
about ninja and that if you have one ninja, it's great,
and if you have a hundred ninja, they have about
the same capacity. Is that one ninja would have just
spread out amongst all of them, right right, That's why
(05:03):
the tick can just breathe right through a horde of ninja. Yeah. Yeah,
besides the fact that he's also ni invulnerable. Um yeah,
they the this is that's exactly right. Yeah. That you
build a robot that you want to be able to
do lots of things. Usually it can do a few
things and not so great. And this is you know,
anyone who's who's played with any of the robots that
(05:26):
are social bots that are meant to do like these things,
like mostly in the toy department, you know, they're amusing,
but they're not exactly the best at anything that they're
trying to do, Okay, And there's solid design reasons for
creating robots that are that that only have one job, right,
I mean, creating robots is kind of expensive and time consuming,
(05:47):
and so if you're trying to build lots of different
functionality into them, that's that's a lot more expensive and
a lot more time consuming if you if you're narrowing
down what you wanted to do, right, if you if
you pinpoint that focus, then you say, here's what the
robot has to be able to do. Anything that does
not directly contribute to this task, we can eliminate. That
(06:09):
means that you can, you know, really pinpoint that focus,
really make sure that that design is as efficient as
possible at doing that task. But let's imagine a scenario. Okay,
let's say that you live in a house by yourself,
and you work long hours, and you need a robot
to help you clean the house. Beyond beyond something like
(06:33):
a roomba that's just going to sweep the floor. Sure,
a roomba might be really good at sucking up dust
off the floor, that's its one job. But a roomba,
of course, cannot clean your gutters, and it can't well
I don't know, I was gonna say, it can't soap
and mop the floor. But they might make rumas that
can do that. They do. They do make mopping room
bas but they're not usually also sweeping room bas So
(06:54):
it's in other words, you have you have two different
robots that have similar but different jobs. Okay, and one
about another one thing. Yeah, another one for washing windows,
and another one for doing your dishes. And there there
have been lots of robots that do these things, but
again they're all individual like their uni taskers, as Alton
Brown would call them. So are you going to fill
your house up with thirty five different kinds of robots?
(07:17):
Well maybe if you can afford it and you don't
mind occasionally getting sharp metal jabbed into you. Well, also,
you just you started this up by saying you live alone.
I mean that maybe they'll make you a little less
lonely to have like thirty five robot friends doing all
your work for you. That would make me more lonely,
would it? Yeah? I just named them. I assume if
you live some with somebody else, you can bully them
into doing all the houses. Is that how it works
(07:39):
in your household? Kidding? My wife is wonderful. Yeah, I
do the same thing, Joe. Uh yeah, so this is
this is exactly the point you're making, though, is is
very much on target right. The idea that it would
be very beneficial to have a mod geiller robot, a
(08:01):
robot made up of smaller robots, smaller pieces that could
form any given shape to perform any task as best
as it possibly could, rather than have a veritable army
of robots. I mean, even if you were to say, well,
let's imagine that I've gotten enough space for all those
robots and I would want them all. Just the power
(08:21):
bill alone to keep them all charged would be ridiculous.
So what if you could have this group of tiny
robots that joined together to form a much larger moving
object that can do pretty much anything you wanted to do? Right, So,
it could form into a small disc and hoover along
the floor like the room but does. Or maybe it
(08:42):
could make itself tall, rearrange itself into a long stick
with arms to clean the windows, or to reach into
the sink to clean the dishes. Or it could even
climb over upon itself up the wall, up the side
of your house to reach the roof and clean out
your gutters. Yeah, yeah, all of these are I mean,
it sounds kind of science fiction e but these are
(09:02):
all real things that, uh that engineers are looking into.
Maybe not necessarily those use cases, but but essentially the
idea of a robot that can change its shape in
order to do different things. Whether that's just to get
across a landscape or to actually perform a specific task. Right, Well,
I think we should talk about some of the existing
(09:23):
models of reconfiguring or self reconfiguring modular robots that exist
out there today, right because obviously we don't have anything
like what we just described, this perfect self reconfiguring robot
in your house, but we we we've basically got a
few steps in that direction that are very encouraging. Yeah,
we've got to that we're specifically going to talk about,
(09:45):
one of which is a reconfiguring robot, one which doesn't
necessarily reconfigure, like it doesn't form a larger robot, but
it can act with a swarm intelligence, which it will
be important in modular robotics. So we're gonna talk about those.
But I should point out there are lots of different
types of modular robots out there. Some of them are
not self reconfiguring. Some of them are simply you get
(10:07):
all the different components and you put them together. Yeah
you can. You can sit there and make you know.
There are a lot of ones that teach children the
basics of robotics, and they have different units that do
different things, and the way you put them together determines
their function. Uh, these are modular, but they don't self reconfigure.
They don't they don't manage to put themselves into whatever shape.
(10:29):
You are the one responsible for doing it. So we're
not really going to talk about those. Okay, Well let's
look at M I T S M blocks. Yes, they
are momentum driven magnetic modular robots. That's how the commercial
would go. You know, I imagine the guy who did
the micro machines ads doing it. Uh no, no, I know.
How the commercial would go would be like, let M
(10:51):
into your life. The M block, you know, M I
T you can just send the checks are away. So
they're cubes that they you would think because they're called blocks,
Joe's already regretting absolutely going to stay in. Okay, So
(11:12):
the M blocks are these blocks that have magnetic corners, right,
so each corner has a beveled magnet in it, which
is pretty cool. I'll talk a little bit more about
that in a second. They also contain a rotating flywheel
that can move up to twenty thousand revolutions per minute.
It's pretty fast, and so, uh, they don't necessarily speed
(11:34):
turn at full speed every time, but depending on what
you need the cube to do you might have it
speeding spinning that quickly, and then what you do is
you apply a break. Now, the laws of momentum state
that that energy has to go somewhere. It gets transferred
to the cube, and this makes the cube move. Okay,
so what does this actually look like when we're watching
these cubes move around on each other? Have you ever
(11:56):
seen a ghost movie where objects seem to be moving
on their own accord with nothing obviously moving them forward. Yeah,
that's kind of what it looks like. It looks like
the blocks are possessed, kind of like kind of like
the toaster jumping and Ghostbusters too. Yeah, yeah, exactly. So
I don't know if you already said this or not,
but they don't have external moving parts. No, it's all
(12:18):
it's all kept inside. The flywheel is actually contained within
the cube, right, So the momentum can move one cube
onto another cube and reposition it around the outside of
that cube without attaching and reattaching via clamps or hooks
or anything like that. But it's it's attachment through those magnets, right. Yeah,
And so if you imagine one it's just one cube
(12:39):
moving around another. But if you add a third, you
could have two cubes moving around on each other, and
then the fourth or fifth, and eventually you can end
up with a large mass of moving modules that can
assume shapes, especially if they're in communication with each other.
And you can also include modules that don't necessarily have
the flywheel inside them. They might have some other UH
(13:02):
specific sensor or camera in them. Uh. The m I
T Group has talked about the possibility of that as well,
in which case the ones that have the flywheel would
be responsible for moving the other ones, which they can
do because that flywheel spinning so quickly. Um, these things
can actually jump, so they don't just like the the
videos I've watched show the cubes. I guess rolling is
(13:23):
the best way of putting it in order to move
across the floor, but they can even Let's say you've
got two cubes stacked already vertically and you have a
third cube roll up. It can actually spin and break
in such a way to jump up and land on
top of those two cubes, which is kind of spooky
looking when you watch it the first time. They also,
(13:43):
of course are making pretty loud noises because it's a
flywheel spinning, and then breaking, so it's it's not silent,
it's it's loud and spooky, so but it's pretty cool.
And and those those magnets that are in those beveled
points in the corners, they rotate. And the reason by
the magnets rotate is so that you can always get
a north south connection. Remember, you know, the opposite poles
(14:05):
attract and similar poles repel, and you want to be
able to make sure that when one block approaches another block,
it's able to actually latch on properly. So the magnets
in those individual points will naturally rotate so it will
be a north south connection so that any two cubes
from any two sides can attach exactly. And if you
get two poles that are the same, then one of
(14:27):
them will end up rotating. Everything will be fine. So
there's literally no wrong way to put them together. Neat. Yeah. Yeah.
These also remind me of another project out of m
I T when we've talked a little bit about before, previously,
before previously, which is the common way to really over
emphasize the point um. They're the m I T S
(14:48):
reconfiguring robots from their self assembly lab um while researching
what's called forty printing. Um. That's three D printing of
materials that can then move and change with with minimal
stimul lists or instruction from the outside. Uh. The kids
of the Self Assembly Lab created some macro scale robots
to help demonstrate the usefulness and honestly the coolness of
(15:09):
this changeability of of structure and so so these are
these reconfiguring robots are programmable folding chains constructed of of nodes,
each of which is hardwired to one or more neighboring nodes.
So if you give one note an instruction, it will
pass information on how to move down through the chain
(15:30):
until the programmers desired shape is is complete with with
relatively minimal input from from that programmer. Right. If you
do not remember us talking about it, or or if
you want to refresh yourself, you can learn more about
these and other self assembly techniques in our video and podcast,
both of which published August. That's like over a year ago. Guys,
(15:52):
we've been doing this a minute. Um. The episodes are
called the video is forty Printing is the Future of Design,
and the pod cast episode is forty Printing is One
D Better. I must have titled that one, actually, I
think I came up with that line. Strangely enough, Yeah,
very it's a very Jonathan kind of joke. So the
(16:15):
other one we wanted to talk about was Harvard's kill abots. Now,
this is the one that is not modular robot but
they do demonstrate swarm behavior, which again is going to
be really important if you're talking about ultimately, if your
if your goal is to create a robot, a large
robot made up of smaller robots, then you have to
build in some form of intelligence for each of those
(16:36):
smaller robots to know quote unquote its place and it's
uh and it's function right. And so there are some
similar concepts going on between these two uh, these two
ideas swarm robotics and the modular robotics. So with a
self reconfiguring modular robout, we're generally imagining something that's connected
to itself. It's got pieces that may all be the
(16:58):
same kind of peace moving on one another to reshape itself.
With swarm robotics, you've still got lots of pieces working together,
but they're sort of moving on mass and they might
not be building a shape, right, they might they often
aren't physically connected together at all, right, they're all individually
free to move. If if somehow one was to break
(17:20):
free of its programming, it can make a break for it. Uh,
it's not. It's not chained physically in any way. So
whereas the ones from M I T were magnetic and
linked together, and we know of a lot of others
that have physical connections of some sort, either they have
connectors that snap in or they have clamps that hold
onto one another, these do not. These are simply to
(17:42):
allow UH computer scientists to test various algorithms for robotic intelligence,
for swarm intelligence. UH and a lot of them are
looking into bio mimicry, the idea of of how animals
and swarms behave in a mass. You know, it's not
it's it's really fascinating stud At any rate. The problem
(18:02):
with these computer algorithms was that for a very long time,
there wasn't an easy way to test them. Like you
might come up with what you think is a brilliant
algorithm to to inform a swarm of robots how to
move for any given purpose, and you can test that
through computer simulation to your heart's content, which is not ideal, right,
(18:23):
I mean, because you're depending upon the fact that the
computer simulation has to be accurate enough to actually reflect
what a physical object would do. So Harvard scientists came
up with this idea and professors as well came up
with this idea to create a relatively inexpensive tiny robot
on which you could test these algorithms by by assembling
(18:44):
enough of them to do whatever it is you need
to do. And they look like, well, they're little circles
like think of about the size of a quarter. But
they are on long, spiny legs and there's a little motor,
a vibrating motor on them that depending on how it's
vibrates the assuming that this is on a level, smooth surface,
(19:05):
the robots will skitter across the ground. Now, their their
legs aren't moving, right, it's just through this vibration, so
it's not like they're segmented legs. It's not that much
like an insect. But they do jitter so that they
move in a specific direction. Yeah, well, these things certainly
do that, And there's some great videos online of these
(19:27):
moving around in various ways. Usually it's a direct overhead shot,
so you just see a little it looks like a
hockey puck, just kind of slowly moving across an area,
so you can even make these yourself if you really
wanted to. I'll tell you more about that in a second.
But the way the way the kilbots form that shape
is all dependent upon the algorithm. Assuming that the algorithm
(19:48):
was was properly constructed so that the robots know what
to do quote unquote um, that they are able to
detect one another and move according to whatever the plan was,
and they communicate through various means like there's uh. The
version I saw, they were specifically using infrared emitters and
sensors so that they would be able to tell how
(20:09):
far apart they were from their neighbors to make sure
they maintained whatever the distance was that the algorithm dictated.
So one of the examples I saw was a follow
the leader um example, where one was designated the leader.
They had actually put a green sticker on top of
it so it made it easy to follow on the video,
and then the next kill abot had the instruction to
(20:31):
try and maintain the same distance with the first one.
The first one's instructions were move wherever you want, but
you can't get too far away from the second one.
The third one's instructions were tried to keep up with
the second one, but don't get too far away from
the fourth one, etcetera, etcetera. So the longer the chain is,
each one essentially has the same instruction maintain as as
(20:53):
you know, as as a stable a distance as you
can to the one in front of you without getting
too far away from the one behind you, and it
would follow the leader this way. And it was a
cute little demonstration. They also had a foraging demonstration where
they had one kilobot designed to look like uh, well,
it just had a red sticker on it. Those designed
to act as the nest this is home base. And
(21:15):
they had another kill abot that was pound off maybe
to three ft away, that had a green sticker that
represented food. So if they wanted to have these robots
go out and search for So we're teaching robots to
be cannibals. Well, it's much better than teaching them to
hate humans. I'm not worried about the killers. Things skitter
(21:38):
at me all they want. Yeah, they're only called killer bots.
And how one letter off from kill bots? Yeah, so
uh yeah, and and well, at any rate, so the
way this would work is that the once they got
the instruction, they used their infrared sensors to look for
the quote unquote food kill and once they found it,
(21:59):
then they would travel back and forth between the food
one and the next one. Uh. Some would just end
up forming kind of a structure around the nest. Uh.
Once they were unable to find the food, they just
kind of stayed still. They wouldn't continuously move around, they
would only go so far away. And that was a
kind of cool demonstration as well. Now, like I said earlier,
(22:21):
you can actually get the plans on how to build
one of these yourself, or lots of them, since one
would not be terribly useful. You can't really test swarm
intelligence on a single robot. The it's difficulty. Yeah, all right,
just pretend you're a lot of people. Uh. But the
whole design is published under Creative Commons. It's just a
non commercial commercial license, so you can't you know, you
(22:42):
obviously can't make them to sell them, but for educational
purposes you can make as many as you can afford
to build. And I say afford to build, because obviously
you have to buy all the hardware to put these
things together. Though, you know, if you're building them in bulk.
Obviously the cost will go down. Yeah, although it's still
pretty expensive because if you were to build say about
a hundred, they price it out as being about fifty
(23:04):
dollars per robot. It's pretty expensive. If you were to
build a thousand, it's twenty dollars per robot, But that's
twenty tho bucks when you take that all into consideration,
and I don't have twenty tho dollars to build my
robot army yet, so uh but at any rate, again,
these kind of test out that idea of swarm intelligence.
More recent videos have shown them do things like assume
(23:25):
specific shapes, like they give it a command to for
the all the robots and I think there were in
this case more than a thousand of them in the
demonstration I saw, but to form a shape like a star.
And so the robots have to have information about which
other robots they're supposed to be next to and how
(23:45):
they're supposed to be configured based on that, which is
really fascinating. You think, you know, this thing is again
about the size of a quarter um. I mean, obviously
it's thicker than a quarter is, but still they're able
to pack in enough of the microprocessor technology for the
robots to be able to determine, oh when when they
we make this shape, I'm number fourty seven. I have
(24:06):
to make sure I'm next to four and three, six
and four and thirty eight. And they do kind of
go one at a time to make this shape you
like it was it was like in a line, as
opposed to all right and break didn't They didn't do
it like that. But it was really interesting stuff. And again,
this is the kind of behavior we're going to have
to see for something like the modular robot that's in
(24:27):
Big Hero six to become a reality. Well, I think
this stuff is awesome, and I want to know what's
currently in development. I don't know if you've looked into that, Jonathan.
I looked into sort of the most recent literature, because
this is one of those things where we still have
people working on implementations. No one, I don't think anyone
has come up with the ideal implementation yet. It's it's
(24:50):
of course not I mean, we're in the we're in
the research phase. Yeah, exactly. This This is a developmental era,
which kind of makes it even more exciting really, because
you know, there's there's so many people looking at different
ways of doing this in different tasks that such robots
could do. That it really starts to fire your imagination.
So for example, here's here just some titles of some
(25:12):
recent scholarly articles that are on the subject. Uh and
and they kind of demonstrate now and like by recent,
I mean these were all published in two thousand fourteen. Uh,
these kind of demonstrate the the breadth of the research
design of modular robot system for maintenance tasks and hazardous
facilities and environments. Well that that kind of you know,
(25:35):
you can easily imagine having a modular self reconfiguring modular
robot would be incredibly useful in that source of situation.
Or how about room bots. A hardware perspective on three
D self reconfiguration and locomotion with homogeneous modular robot. So
here we have a specific scholarly look into well this
is you know, we we know what we want to achieve.
(25:58):
What are the way is to do it? Using this
specific approach where you've got robots that are all essentially
the same shape that combined with one another, then we
have design principles and testing of a latching modular robot
connector So again looking at different ways for these different
modules to join up exactly. H then you know the
(26:19):
modular robot joint design and experienced verification for small openings
and enclosed space, which again really awesome because you're you're
looking at the possibility of using these robots that could
go into places that you know, if it had a
solid form factor that didn't change, it couldn't get there.
So imagine that you have to send uh, this robot
(26:40):
into a building that is you know, perhaps it's a
really old building and it's got U faulty structure, you know,
structural damage, and you want to make sure it can
navigate all the way through and actually investigate that structural damage.
That could be really useful. Sure, yeah, if there's a
collapsed path or something like that and you need to
get under it or round it or whatever that it is,
(27:01):
that that is right right, being able to break down
into all of its creepy little robot parts and skirt
her on through is terrifically useful. You know, it's stasis
on the terror. You know. It strikes me that there
are sort of like two main ways that this technology
can move forward, and really both of them seem crucial,
but one of them is about working on the modules themselves,
(27:24):
like making each individual piece, and we're assuming we're working
with robots that are made of pieces that are all
identical pretty much, Uh, that making each individual piece cost
effective but very versatile, able to move around, sturdy, able
to communicate, without being a gazillion dollars apiece. But then
also talking about the overall swarm intelligence, how do you
(27:48):
program the robot through these modules to have some kind
of collective brain that enables it to do its work effectively? Right? Right? Yeah?
And this is I mean, these are a lot of
unanswered questions right now, right can could there be an
emergent artificial intelligence from the swarm intelligence that would allow
(28:08):
you to essentially tell it here's what I need you
to do, you figure out how to do it. Or
would you always need to have some sort of controlling
system that would essentially send instructions to that swarm of
robots to form a particular shape to complete that test.
Oh yeah, that's an interesting approach. I hadn't even thought
(28:29):
about that because I was just imagining we're assuming this
has to be autonomous in one way or another, but
you could just I guess, have a laptop that's bluetoothed
up to your modular robots swarm, and this thing is
reconfiguring itself to do work based on orders that are
sent to it from the software on your computer. Yeah. Well,
I mean when it comes down to it, the question
(28:51):
is how autonomous does it get? Because there's some autonomy
that is necessary, right, because if you're talking about thousands
of robots, you don't have the time to tell every
single individual robot where it fits in this grand scheme. Sure. Sure, well,
I mean you know, if you think about something like
like having a a lead butt with a green sticker
on it or what have you, um, that you can
bluetooth up with and give a command too, and then
(29:13):
have it break that down into commands to give to
everyone else in the swarms that you don't have to
literally contact every single microbot, right. I think there will
be a bridging system. Yeah, I was assuming the commands
would be given by software, not like manually by well well, sure,
I mean you know, no EI either way though, Um,
there are lots of different factors of trying to communicate
(29:34):
with a swarm like that, and how you would go
about making it the most efficient, and it's kind of
like when we talked about neural networks, how every time
you add a node, it complicates the entire structure. I
don't want to say exponentially, but quite a bit large fold.
So same sort of thing with modular robots. The more
modular robots that are in the system, the more complex
(29:56):
it's going to be to command it. But that's what
you know that swarm and eligence is all about. So
that's really exciting and uh, you know, kind of looking
into the future, like let's assume again kind of like
Big Hero six did. Let's assume that this this research
pans out and that we come up with some implementation.
It doesn't you know which which one is immaterial for
(30:18):
this discussion, but some implementation that works. And then in
the case of Big Hero six, Uh, it was all
magnetics that held everything together, and the individual modules look
like a sphere with two little uh protrusions that stuck
out that could that were a kind of hinge. They
could move independently of each other. So uh, that's what
(30:39):
the individual component was, and then together they could form
pretty much any shape you you wanted. Um, and they
did so using a brain computer interface, which was really cool.
Is essentially a headband you could wear where your thoughts
get transmitted into computer commands that tell the robots what
it needs, what they need to do. Uh, that's awesome. Anyway,
(31:04):
we're quite a ways away from that too, But people
are working on brain computer interfaces just as they're working
on this modular robot issue. So you would want, in
an ideal implementation to have your microbots as small as
you could possibly make them, right, because that that would
increase their versatility. But also, you know, you can think
of it kind of like an image of digital image
(31:26):
and in pixels. You know, if you have smaller pixels,
you have better resolution. The picture looks better as opposed
to large blocks of color than it looks blocky and
and jagged. Same sort of thing with your modular robots.
The smaller the unit, the more smooth the robot's going
to be, the more versatile it's gonna be. Um, but
that also comes with some pretty big issue problems. But
(31:48):
another one is that this ideal implementation would allow it
to move through those small spaces we were talking about,
whether individually, you know, so it splits apart and all
the individual pieces creep through an area, or it forms
some other shape, like a snakelike shape to get through
certain areas. I've actually seen some really awesome snake bots
that are incredibly disturbing to watch on video. Uh, we'll
(32:12):
probably need to power these suckers with something that we
haven't thought up yet. Yeah, this is a real problem.
Oh sure, you know, well, any significant pushing or pulling
or lifting or other interaction requires really a lot of
power right now. Hydraulics are really the most efficient systems
for for that kind of stuff, and they're you know,
for example, what what drive the most badass exo suits
(32:35):
for for lack of a better term, um, but they're
also really clunky and don't really lend themselves to this
microbot kind of format because you need like a tank
of fluid in order to drive. I can't imagine having
like super small lines to each microbot and still having
any kind of degree of freedom. Yeah, it would be problematic.
Yeah yeah, so I mean, you know, maybe the answer
(32:55):
to this is that we were not going to use
this kind of system for any kind of heavy lifting. Um.
But even so, just powering small things that you want
to keep lightweight is a problem. All the time. We
talk about it always add weight. So one other challenge,
and we've talked about this, is the fact that we
need to have these these individual units be able to
(33:19):
communicate with each other at least on some basic level
so that they quote unquote know what they're supposed to
do and where they're supposed to go. And that's where
we're seeing this this work with, like in the Harvard
kill abuts of of Swarm Intelligence. It's it's one thing
to see the work that's being done with things like
the the M blocks where you're seeing maybe six or
(33:41):
seven of the units working together. It's another thing when
you start talking about you know, large collections of these devices.
We talked about that a little bit before too, in
our episodes about ants and also about biommicry as a
whole um and also in our upset about right emergent behavior. Yeah,
(34:02):
these are all things that you know, we don't have
answers for because we're still asking the questions. But it's
really exciting again to see the work and while knowing
about the stuff because you know what, we've researched so
much into robots and artificial intelligence just for this show. Uh,
the not not just this episode, but the forward thinking
series overall that when I went in and saw this movie,
(34:25):
when I was watching Big Hero six, That's why I
was like, I had a huge grin on my face
because I was thinking, these are all the things we
talk about on our show. Uh that where we are
now in a world in this movie where all those
things have been realized, where all the problems and challenges
we talk about have been overcome, people have figured out
the way to engineer a system that meets those challenges,
(34:48):
and to look at the potential of that world was phenomenal.
I was like, Ah, it's like I'm watching the outcome
of a forward thinking podcast where we have that bit
where we say, what will the future like? And this
is what the future would be like? And it's it's
an amazing vision of that future. So um yeah. The
first ten minutes of that movie, I was like, man,
(35:10):
I wish they had called me because this is awesome.
It feels like it's the outcome of a forward thinking show.
Um but you know, and then of course it gets
into more fantastical elements as it goes on, and and
the science and technology get a little more loosey goosey,
but they had already won me over at that point,
it didn't matter. So you have a hard heart for
(35:32):
scientific implause. There's I can get pretty nitpicky. I admit
it wouldn't happen like that. Yeah, yeah, well, I mean
they're there. Definitely comes a point in that movie where
you gonna say, all right, I just gotta go along
with the ride Bernell or I want to be it's
it's it's worth seeing you guys should definitely see it.
And of course, you know the Smudgler robot research. We're
(35:55):
going to keep an eye on it. It's fascinating stuff.
We're learning so much about robot intelligence and robot mobility
through this kind of stuff that it's just a fun
thing to keep our eye on. So, guys, if you
have any suggestions for future episodes, maybe there's something else
about robotics you've always wanted to hear about. Obviously we
don't shy away from that topic. Let us know. Or
(36:17):
if there's anything else you know, just something you're wondering about,
what would that be like in the future, send us
a request. We read all of them, we're happy to
get them, and we would love to hear what you think.
Send us an email addresses FW thinking at how Stuff
Works dot Com, or drop us a line on social media.
We are on Twitter, Google Plus, and Facebook. At Twitter
and Google Plus, we used to handle f W thinking.
(36:38):
Just search fw thinking and Facebook. We'll pop right up.
Leave us a message, and we will talk to you
again really soon. For more on this topic in the
future of technology, visit forward thinking dot com, brought to
(37:02):
you by Toyota Let's Go Places,
Fw:Thinking News
Hosts And Creators
Jonathan Strickland
Joe McCormick
Lauren Vogelbaum
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.