A new revolution

Lecture 3 - Sparks will fly: How to hack your home

Making music

Inspired by the Royal Institution’s very own Michael Faraday, Prof Danielle George attempts to use simple motors to construct the world’s greatest robot orchestra.

When Michael Faraday demonstrated the first electric motor in 1822, he could never have dreamed that in 2014 we’d be surrounded by mechanical devices capable of performing nearly every human task. In this lecture, Danielle explains how these robotic and motor-driven appliances work and shows how they can adapted to help you kick start a technological revolution. She shows you how to turn a washing machine into a wind turbine, how Lego can solve a Rubik’s Cube and how the next Mars rover will traverse an alien world.

Themes

Engineering

Details

Type:
Christmas Lecture
People:
Professor Danielle George
Location:
London, UK
Filmed in:
The Theatre
Published:
2015
Filmed:
2014
Credits:

Windfall Films / BBC / Royal Institution

Collections with this video:
Sparks Will Fly: How to Hack Your Home

Licence: © Royal Institution

Comments

Transcript

[MUSIC PLAYING]

Tonight, we'll be assembling the world's greatest robot orchestra and explaining how simple motors will allow these machines to perform alongside human musicians. Welcome to "The Christmas Lectures."

[MUSIC PLAYING]

[APPLAUSE]

The world is full of robots, automated machines that use simple motors and clever software to copy the work done by humans. And they're getting smarter all of the time. Take a look at these little fellows here. Robotic acrobats that can balance on a tightrope. Or our six-legged robot spider here that can shift its weight to stay upright, just like a human can.

But what about some things slightly more complicated? What about a musical instrument? I could do this, but I don't think you can do this.

["AW"S FROM AUDIENCE]

Well, we'll see if he can do it later on. Let's see. Now, if you've seen any of my earlier lectures, you'll know that I've been setting myself a grand challenge each time. And tonight is no different. The first person to achieve continuous movement with electricity and demonstrate the first electric motor, which is essential for all robots to function, was the Royal Institution's very own Michael Faraday.

Now he was fascinated with the relationship between electricity and movement. And tonight, I want to honour his work. So let's do something amazing with motors that Michael Faraday could never have imagined in 1821. So drum roll please. Oh, good.

Tonight, we are going to construct the world's greatest robot orchestra. Now the sound of an orchestra playing in perfect harmony, to me, represents the pinnacle of human achievement. So the question is, can robots replicate this? Over the past few months, engineers from across Europe have been building robot musicians. And tonight, we've set them the ultimate challenge-- to play the "Doctor Who" theme tune.

Now some members of our ensemble have used hack technology from around the home to make the music. Others have programmed and trained to play much more traditional instruments. And they'll all be accompanied by real, living members of the London contemporary orchestra. Honestly, this truly is going to be great.

So let's break down our problem into steps to build up our orchestra. We need to break the tune into its components for the "Doctor Who" theme tune. So first thing we need is the rhythm section, the drums. Then we need that low base synth sound, so we need some synthesisers. Then let's add some guitars into that. But then we need the melody for our "Doctor Who" theme tune, and we also need a keyboard in the tune.

And then, for a bit of fun, let's make one of our instruments fly. Now one thing all of these robots have in common is that they'll all be relying on the simple relationship between electricity, magnetism, and movement. So that's where I'm going to start. This is the world's first motor, and it was demonstrated in this lecture theatre by Michael Faraday over 200 years ago. And Charlotte, who is the curator of collections at the RI has very kindly brought this in for us.

Now it's so fragile that nobody's allowed to touch it apart from Charlotte. So thank you very much for bringing this in, Charlotte.

You're welcome.

So Andy has very kindly built a replica for us as well. Now Faraday knew that if he could run a current through a wire, this would create a magnetic field around that wire. So if we place the wire next to the magnet, with the magnetic field running perpendicular, or at right angles, to the one created around the wire, the wire would move. But the movement would break the circuit, so we couldn't really get continuous movement until he remembered mercury.

But before we get this mercury out, I think we should get the real fragile one out of the way. So thank you very much Charlotte.

[APPLAUSE]

So Andy is going to pour the mercury around the magnet. Then, we can dip this metal needle that Andy has here into the mercury. And we can pass an electric current through it. So we can just use a normal power supply to pass the electric current through it. And then what we should see is rotation.

So the mercury metal conducts electricity, but it also allows the liquid, which at room temperature the mercury is a liquid, so the objects could move freely through that. And there we are. We have rotating. And it would just keep moving round and round, because there is liquid in there which allows that continuous movement.

But of course, in modern motors, we use brushes instead of that mercury. But thank you very much for your help there Andy. Now motors still work on this principle, and you can even try building one yourself. You can make one, actually, out of a battery, a magnet, and a coil of wire. And I'm going to come and sit next to you, if that's all right, to come and do this. So shifty along everybody. That's enough. I'm not that big.

OK. Now this is a really simple one that you're going to help me with. OK, so what's your name?

Lucy.

Lucy, OK. So what we have here Lucy is a battery, a magnet, and a coil. Nice and simple. So what I want you to do is just put that battery on top that magnet. OK. You see it's a very strong magnet. So that magnet can set up a magnetic field. Then we have our coil, here. And if we place our coil over the top of that battery, it will generate a current, an electrical current, through our coil. So the wire sets up a separate magnetic field which runs perpendicular, or at right angles, to our magnetic field here. And the result is that the two fields will react with each other and our coil rotates. OK?

Yeah.

Got it Lucy? So place our coil over the top. There we are.

[APPLAUSE]

Then it goes round, nice and fast. Thank you. Now in a moment, I'm going to be introducing you to the first robot in our orchestra. But just before I do, I want to show you a hack that we did with the most common robot of them all. Now you might not think your washing machine is a robot, but actually, it's just a collection of motors that replicates the manual work of having to wash your clothes by hand.

So on a much more basic level, it turns electricity into movement. So if it's possible to turn the electricity into movement, can we turn movement into electricity? Well, a few weeks ago, I challenged Andy to build a wind turbine and generate electricity using an old washing machine. Now this isn't as stupid as it sounds. When you put your power into your washing machine, you get movement. The drum spins. So if you spin the drum, surely you can get power out. So Andy, how did you get on?

Not too bad, not too bad. I didn't use too many of the bits of the washing machine in the turbine, as you can see, littered around here. But the main bit, obviously, is the motor. This is the main motor that turns the drum in the washing machine. And like you said, if we can use something else to turn the motor, it should generate electricity. So we're going to be trying to use these turbine blades to turn the motor to try and generate some electricity.

OK. How much electricity?

Maybe not all that much. This probably isn't the most efficient way of generating electricity in the world. But we should be able to get some light out of these torches on the front here.

Right, OK. So it's these we need to keep an eye out, is it?

Yep.

Right. So we need to test it. Now of course, it doesn't get that windy in the lecture theatre. So Andy, quite conveniently, has brought his leaf blower with him. So it might get a bit blustery over here, so hang onto your heads, everybody. OK, so are we ready? So everybody keep an eye out on those lights. So let's test it, Andy.

OK.

[MOTOR WHIRRING]

Good, yeah! Fantastic. The blade is spinning, and the lights came on. So we made a wind turbine from a washing machine. Now, as Andy said, it's not the most efficient way to power your home. But it does show that relationship between electricity, magnetism, and movement really well. Excellent work Andy, thank you very much.

You're welcome.

Now we know a little bit more about motors, we can start assembling our robot musicians. So let's start simple, with the drums. Well, maybe not so simple, but you only have to worry about the rhythm. Because on the drums, the pitch doesn't change. So all we really need is a single motor to play each drum. Like this snare drum.

So attached to each drumstick is a motor that will connect the circuit so that the motor makes the stick hit the drum and bounce back again. So I should be able to just keep hitting that drum with every signal, like so. And it worked! Brilliant. But our orchestra isn't going to work by me standing here pressing a switch, so we need to programme that tune.

For a human orchestra, we'd use sheet music, which has a line for the base, the synth and the strings, and of course one for the drums as well. Now sheet music is similar, actually, to a computer programme, in that it tells the musicians what notes to play, how long to hold each note, and how long they should play it for. Now let me show you a robot that uses some very simple instructions to do something very, very cool.

Remember these? Has anyone ever tried solving a Rubik's cube? Oh, lots of people. There you are then, you can have that one. Try solving that. And someone over here? You try solving that one. I need this one, I'm afraid. OK, now, you should have solved it by now, yes? How are we getting on? No? Not quite? OK, well, keep going. Now it might seem impossible to solve, but mathematicians have shown that you can solve any cube within just 20 moves. Because there's a certain pattern of twists and turns that would get the cube back to normal. So all of the colours on the side would be the same.

Now the inventors of this robot set themselves a challenge. This is cube stormer three, and it's the quickest cube solver in the world. So how long do you think it would take to solve this? How long?

10 or 20 seconds.

10 or 20 seconds. Yep. Any other ones?

30 seconds?

10, 20 to 30 seconds. OK. Well, let's see. I just align my Rubik's cube, and if Dave you just want to come and have a look at this. I press my go button. Wow. Four seconds! I know it looks magic, but actually it's LEGO and a smartphone. Now it takes photographs of the cube, and the software works out the quickest way in which to solve these Rubik's cubes. So it writes a programme to tell each robot arm which way to turn. And then actually, once the maths is done, it's just a simple list of instructions. So spin this part of the cube, rotate that part, spin the other face.

Now the same is true for our robot drummer. Now he's from Queen Mary's University of London, and here he is. Say Hi to Mortimer. Hi, Mortimer.

Hi.

Now we need to write a programme to know exactly when each motor will turn it to create the rhythm. Now the easiest way to do this is a form of electronic sheet music called MIDI, or Musical Instrument Digital Interface. And it's being designed specifically so that instruments can talk to the computer and back again. So instead of writing a whole new programme, we can take the drum from our sheet music, write it in MIDI software, which will then convert it into instructions for Mortimer. So if we sent a simple code to the drummer, we should here this.

[CYMBAL PLAYING]

OK. Nice and simple. But we could send a more complex code that repeats a certain rhythm, and we'd hear this.

[DRUMS PLAYING]

I am a drum voice.

[DRUMS PLAYING]

Excellent. OK. Already we're starting to rock, and we just have one. Now, in fact, all our robots are going to be controlled by MIDI, as it means we can work in a format that humans can read, which is sheet music, and then convert it straight into code that our robot musicians are much more comfortable with.

OK. So one musician down, lots more to go. Because we don't just need a rhythm in our piece, we also need pitch. So we need to start with that low base synthesiser sound. You know, the one for "Doctor Who"? Who knows it? Sing it, everybody.

[SINGING]

Not bad! Not bad. I hope the robot orchestra are as good as you guys. So we need another robot that can read music and adjust the pitch of the sound as it produces it. Now in the spirit of making, and hacking, and repurposing, this robot musician is made from a device that was made redundant many years ago.

Now in the days before laser and inject printers, documents were printed by very noisy dot matrix printers, which had to punch a letter shape through the ink-covered ribbon onto paper. Now we don't really think about printers as robots, but they use motors to replicate human work. Or at least, the hundreds of monks that used to copy books.

So if we set it printing, we'll hear the motors whirring away and changing its pitch, but we're not interested in what's being printed. We just want to hear how the sound changes. So the owner of this has hacked it, and he realised that he could alter the pitch of the sound by printing different patterns. So it accepts our MIDI code.

So the computer in the printer convicts into instructions telling different motors when to move. So it sounds like this. Let's see if you can guess the tune.

[PRINTER CREATING MUSIC SOUNDS]

Fantastic. Did anyone guess it? Anyone guess it? Yeah?

"Ode to Joy." Well done, yes. And well done if you knew that at home, as well. Now our orchestra is starting to take shape. Robots can convert MIDI into motorised movements. And there's a much more sophisticated breed of printer that's starting to increase in popularity and may prove very useful as we move to the next instrument in our orchestra.

Finishing off the rhythm section of our unconventional orchestra is a bass guitar, which has been adapted to play itself the students at the University of Leeds. So like our earlier instruments, it reads our MIDI programme, but this uses solenoids and compressed air to press down on the strings, and then it adjusts the pitch of the notes played when the actuator plucks the strings, so just like a human guitarist would.

So let's hear what our bass guitar sounds like.

[GUITAR PLAYING]

Wow. Really interesting. But the team who put this together wanted to give the guitar an even more human like feeling, so it gave it fingers. Now the solution for these fingers was to print them using a 3D printer, which is quite similar to this one that we have.

Now we have to be very careful with this, because it's halfway through making a model. And it's making a model of someone in the audience. So Isla would you like to come and join me? Hi, Ila. How are you? OK. So let's just go around here so we can see what's going on, Isla?

So a few weeks ago, we sent some people to Isla's house to get a 3D scan of her head. And now-- and we can see the people scanning your head there. That must have been very strange for you, Isla, was it?

Yeah.

So the head is being printed layer by layer. So each of these layers can be seen as a thinly sliced horizontal cross section of Isla's head. Now 3D printing is just a rapid prototyping process which is capable of making three dimensional solid objects from a digital file. Now to buy a 3D printer is actually still quite expensive. But if you have a design, there are companies that will print one for you, just in the same way people print your photographs.

So as you can see, this would take a very long time, to start printing Isla's head now. So we have printed one a little earlier. So are you ready for this, Isla? Wow! What do you think, Isla?

Could I get you standing together with your profile? Do you see that profile there? I think that looks great. What do you think, Isla?

I think it's good, yeah.

It looks fantastic.

I did have my hair up at the time.

You did have a ponytail in at the time, did you? Good, we're glad. Well, the extra special thing here is that your bust is going to go into the Royal Institution downstairs, next to Michael Faraday's bust. How good is that? There you go. All right, thank you very much Isla. Thank you.

Now printing colourful fingers is fun for our bass guitar. But is there a more serious application for our 3D printing? Well, yes. And hopefully, we have a video call here with Hayley Fraser in Inverness. Now, hi, Hayley. How are you?

Fine.

Hi.

Good, good. Now Hayley, I understand you have a prosthetic hand that's been 3D printed. Is that right?

Yeah.

Yeah? Can you show us it? Fantastic. Do you want to give us a wave? That's brilliant. Now, this must be life changing for you. Is it?

Yeah. I really like the colour, Hayley. The pink colour. Did you choose that?

Yeah.

You did, yes? OK. Now can you show us something with the hand, maybe pick something up?

We'll let Barbie see that.

So Hayley's dad, when did Haley get this hand?

Hayley received it back in June this year. It took around six weeks for the whole process, which was a very quick turnaround.

Wow, that's amazing. It must have made Hayley's life just so different, did it?

Oh yes, without a doubt. Yeah.

Yeah. I think the thing that really impresses me most about this is that prosthetic hands can cost literally thousands of pounds. But this one is so low cost that it means that, as Hayley grows, then the hand can always be the right size for Hayley as well.

Yeah, mhm.

And that must be life changing in itself.

Yeah, of course. We can actually buy 3D printers for our home, which we can actually print the parts for herself.

That's fantastic. So you can actually print your own hands at home then, as well?

Yes, yeah.

So it means you can choose what colour you want as well, Haley. Yeah? That's good. OK, well thank you very much for joining us, and happy holidays to you.

Thank you very much.

Thank you, buh-bye. Now our orchestra is starting to take shape. We can have a drummer, a bassist, and a printer. Let's not forget our lovely printer. While these robot musicians will get away with playing the right notes in the right order, some instruments are much, much harder to play. Now, we want a robot to play the solo part in "Doctor Who." Who knows the solo part in "Doctor Who"? Sing it.

[SINGING]

Someone down here is very, very good. Excellent. If it all goes wrong later, I might just get you to do that, OK? Now we want them to play it on one of these, a theremin. And they are notoriously difficult to play. Because you have to play in midair.

[THEREMIN PLAYING]

So you can hear some sort of strange noise. And on the left hand, I can control the volume.

[THEREMIN PLAYING]

And then my right hand can control the pitch.

[THEREMIN PLAYING]

Well I've definitely not perfected it yet, so I hope the robot does a lot better than me. But one of the things that makes this especially hard to play is that to keep in tune, you have to position your hands at exactly the right place in the air. If the theremin is nudged or the settings are different, you'd have to reposition your hands. So you have to listen to the sound that's being produced, think whether it's in tune, and then think, do I need to pitch higher or lower, and then move your hands accordingly.

Now this is called a feedback loop, and this is what our theremin-playing robot will need to master. Now to show you what I mean about a feedback loop, I need a volunteer to help me with the swannee whistle. OK, you there with the necklace on?

Yeah.

OK, come down.

[APPLAUSE]

What's your name?

Alexie.

Alexie. OK, Alexie, right. What I want you to do, here's your swannee whistle. I'm going to turn away from you and play a note. And then I want you to try and replicate that note. OK?

[PLAYING SWANNEE]

Got it?

[PLAYING SWANNEE]

Keep going.

[PLAYING SWANNEE]

You're nearly there.

[SWANNEE PLAYING]

Oh, I think you just about got it there. But it's quite difficult, isn't it? But actually what you were doing there was a feedback loop. So I gave you a note, you were trying to listen to what that note was, and then you were adjusting your hands and the sound accordingly so you hit to same note as me. But it's quite difficult, isn't it? Yeah. You did very well. Thank you very much.

[APPLAUSE]

Now this rather cute little robot is programmed to follow a very simple feedback loop. So when I set it running, what we need is for it to follow a white line. So we just have some white line on a tape around a theatre. Now the programme we've written is a nice simple one. So if I set this running on the line, we can see it start to move. So the programme is written. It's very simple. At the front of that robot is an infrared light and a sensor. Now when the light flashes, the amount of light reflected back is measured by that sensor. So if that sensor is over the white line, that's a high value, and the computer tells the motor to move the robot forward. If it's over the black, the amount of light reflected is too low. So the computer tells the motors to move the robot left or right and find that white line again.

And each of these is a feedback loop. We have an action, the robot moves. That's a measurement with the sensor, the robot reacts, and it changes what it's doing. And you can even see, it's just gone around a little corner there. So I think we might just let it keep going. What do you think? Let's see how far it goes? OK. Well we've got a camera mounted to the front of it, so we'll just follow its progress and we'll pick up with it later on.

Now more advanced robots, like the one capable of playing a theremin, can't really be programmed in this way. Our little line following robot knows what it's likely to encounter, so we can tell it exactly how to react. But what happens if you place a robot in an unpredictable, alien world, without a big white line to follow, in a place where communication is almost impossible and robots have to deal with the challenges themselves?

So on the surface of Mars, for example. So please give a very warm welcome to Abbie Hutty and Bruno.

Hi, Abbie. Thank you very much for coming here. So Abbie, you've heard about our challenge to make the most advanced robot play a theremin.

Yeah.

Now you're part of a team that's building a much, much more complicated robot. So what do we have here?

This is Bruno. This is one of our prototype rovers for the 2018 Mars Rover. So that's the European Space Agency's first robot mission to Mars. So Bruno here is a prototype. That means he is a working model. And he helps us to develop things like how we're going to drive around on Mars, and also how we're going to actually autonomously navigate. So how are we going to work out where we want to travel when we're on Mars?

Fantastic. Now I know tonight that Paul is actually controlling him for us in the lecture theatre, but I'm guessing that's not what's going to happen when he's on Mars.

Yep. So in the lecture theatre, it's pretty dark in here. And the rover is designed to be able to see and get the right kind of contrasts on Mars during the daytime. So he can't really work out things in the dark. So we're actually having to control him by Wi-Fi. Which is great. We're just remote controlling him. But we can't do that when we're on Mars, because Mars is so far away that it would take up to 22 minutes for the signal to get there, just to tell them what to do. So if you're driving around on Mars, and you hit the Stop button, that's a bit too long, clearly.

Yeah, absolutely.

So we have to make our rover intelligent enough to make its own decisions about what it can and can't do safely and drive around all by itself.

Brilliant. So you're almost like giving it coordinates that you might give a car GPS?

Yeah, it's a bit like that. We can give it a destination that it's got to go to. It can be up to two days drive, and it'll just look at what's in front of it. It can bring up an elevation map, which is basically a picture of what's in front of it in three dimensions. And then it can do calculations to work out what's too big an obstacle to climb over and what's safe for it to trundle over, and then just pick its own roofs to that destination, phone us up.

I think we have that here, yeah, You can see.

So this is an elevation map. So this shows us there's a big rock in front of us here that would be too big to climb over. But all of the blue area, that's nice and safe. Nothing's too big. So we can just drive straight over that.

Right, OK. So the other thing I'm not thinking Abbie is that, these wheels, they don't have the normal rubber on them, like you would on tires. So they're all metal, so they're very noisy. Why is that?

OK. So the primary objective of the rovery mission is to look for life on Mars. Either in the past, or present, living life. And rubber comes from trees. It's a natural substance. So if we were to take rubber tyres with us, that could contaminate the samples that we're looking at. And we could find Earth life in that rubber and think that it was life on Mars. So we've got to make sure that nothing that we take with us is going to contaminate Mars. We've got to develop these flexible wheels, so that you still get the same kind of traction and grip going over rocks and going through sand as you would with a rubber tyre, but without anything organic inside it.

Wow. OK, that sounds like a tremendous project. Thank you, and all of this technology that you're developing here will be on the actual Mars rover. Is that right?

Absolutely. Yeah, so we're developing everything in bits and pieces. Lots of different prototypes that test different things. And it will come together in our final rover that launches in 2018.

That's fantastic. Well thank you so much, Abbie, for bringing this along. And the very best of luck with this fantastic project.

Thank you.

Thank you.

That, that was incredible. Absolutely incredible. And to play our theremin in our orchestra, our robot will need the skills of Bruno, that Mars rover. Because Abbie didn't tell Bruno exactly how to get to its destination, she just tells him where to go, and he was working out the rest. So in the same way, we need to be able to tell our theremin player which note to hit, and it must make a judgement as to what the note might be.

Then it listens to the sound that's being produced and makes adjustments to find that perfect pitch, just as we were doing with the swannee whistles earlier. And here is our solution. We are back with our H5W. It's such a cute machine, isn't it? It's modelled on a three-year-old, and it lives in Barcelona in Spain, and it's travelled here together with several friends, including Paul Verschure and Vicki Vouloutsi from SPECS Research Group at the Catalan Institute of Advanced Studies at the University Pompeu Fabra. Welcome.

Hi, H5W.

We have a happy robot.

My name is H5W. I am iCub robot. Paul is my friend, and we love to be here.

Brilliant. Now Paul, why would you need such a complicated robot?

Well the iCub robot, this one called H5W, has about 60 motors. Like many of the robots we saw can do one thing. But if you look at their own bodies, we could do more than one thing. On the other hand, if you want to really advance robots, and make a robot that really can be our friend, that can work with us, and we can relate to, then we might have to think about giving them more of the properties that we have.

And it does feel really human like. When it turns around and looks at you, and winks at you, and things, it does feel very, very human like. So let's see what it can do with these hands. So H5W, can you do something?

Let's play with the piano.

That sounds like a good idea. OK, so why don't you play a C? Can you play a C for us?

Well, that was almost right. And we have been practicing this the whole afternoon, you know, so I'm a little bit disappointed.

It's like a naughty three-year-old.

Look, H5W, can we try this again? And keep in mind, it's being recorded. Lots of people are watching. Can you try to do it right this time? OK.

Well done.

Thank you very much. That was a lot better.

I must practice more.

Yes, I agree with that.

So we've been talking about feedback loops, and how we need that for our orchestra. So this is a very sophisticated feedback loop plus the one step further from that as well, isn't it?

Well this is the whole point. If you look at our brain, it's actually managing different kinds of feedback loops at the same time. You can think about brains like a prediction machine. So it's not only getting errors from the world, as in the line following robot. Oh, I'm off, now I have to correct. It's like a feedback error. But it's really predicting. Like, oh, what should the world look like when I'm doing things? So it's highly relying-- same for this robot-- on the predictions it's making about the world it's in.

And we'd need that if we were able to coexist with robots.

Absolutely.

So we'd like to hear a tune from it. I think we'd like to hear a tune, wouldn't we?

Yes.

All right. Let's see what it can play.

This was not really that difficult. How about this? Could you guess that sound?

What do we think? I think that was starting to sound like something. What do you think it was?

"Twinkle, Twinkle Little Star." It sounded a bit like that.

Very good.

Yeah, excellent.

Which is correct. Well done.

It's correct, well done. So it sounds like H5W could be the star of our theremin playing later on. So thank you very much for showing this, Paul, and thank you very much H5W.

[APPLAUSE]

Thank you.

So I'm looking forward to hearing H5W play our solo part on the theremin later on. Now by using constant feedback loops, we're now one step closer to completing our orchestra. So how is our line following robot getting on? Can we see any-- oh yeah, we can see some footage of him. Excellent. OK. So you can see him actually going around quite a tight corner there. That's quite difficult to do. So he's definitely outside of the lecture theatre, and somewhere in the RI there, I think. But I think we should just put some tape down outside the RI and just let him walk around London, see what's going on.

Now next, we want to work out how to bring several robots to play together. After all, a good orchestra is much more than the sum of its parts. Now to give you an idea of how robots work together, I want to introduce you to a swarm of robots. These rather cute little things here are pixel bots, and they're being developed by Disney research and ETH Zurich. And they're tiny, two-wheeled robots that have LEDs so that they can light up and make shapes.

So to tell me more about these, please welcome developer Paul Beardsley. Hi, Paul. Hi. So thank you very much for joining us, Paul. Now these pixel bots have started swarming together in the middle, but they're going to start a performance for us, I understand. Is that right?

That's right. This is a swarm of robots that can make images and animations. And what we're about to see is the story of the universe, as told by the pixel box, all in two minutes. So the way to think of it is each robot is like one pixel. And what we've got here is a display with 50 pixels.

OK. So we're seeing-- is something happening here, Paul? What's going on?

So now, we're seeing the story of the universe told by the pixel box. And it started with the Big Bang, and now we've gone to the solar system, an abbreviated version. Here's the sun, here's the Earth, and here's the moon. And what we're going to see is a fish appear, and then a dinosaur, and then a human, and that's the history of the universe in an abbreviated form.

History of the universe by pixel bots. Brilliant. And do they-- I see there's a couple of them colliding. Do they often collides?

Well, this is one of the things we've worked on, is collision avoidance. So if you've got a swarm of robots, you want them to go out into the world. You don't want them to collide with each other or with people. These are little robots, and so we forgive them if they make a few mistakes. So you will see a few collisions happen.

Yes, absolutely. So how are they communicating? I notice you've got some antennas just over here, as well.

Yes. That's right. So the way this system works is there's a camera above here, and that's connected to this computer. And the computer, it knows the images we want to create. It's also computing the collision avoidance. And it then sends wireless commands here, and they go to each individual robot. So we're saying, robot 1, go here. Robot 2, go here, and so on.

Excellent. And we're seeing a dinosaur here?

That's our dinosaur, yeah.

OK, brilliant. OK. And then of course, evolution will give us man at the end here.

Of course. One step beyond these robots. And the image will--

OK, well, I think I'd quite like to get a volunteer. So how about a volunteer? How about you there with the grey jumper? Yeah. OK, what's your name?

Jocelyn.

Jocelyn. OK, Jocelyn. What I want you to do is try picking up part of the body. OK? A pixel bot. And pretending it's a naughty pixel bot, put it over there. Just move it out of the way. OK. Let's take a leg. See what happens. So you can see, they're actually all compensating for it. So that's now the leg has just become the hand. Let's try and trick it Jocelyn. Let's take two or three. So move a couple of the legs, and a body. Get a green one as well. Yeah. Get a foot, as well. Right. And then plunk them down. Let's see how they get on. Wow. This is amazing, Paul. So you can see that they want to move. And then not only that, they can actually change colour. So the foot has now become either a hand or an arm, and the rest of the body has compensated for it.

That's right.

That's amazing. I don't think we can outsmart it, Jocelyn, can we?

Unfortunately not. But that's brilliant. Thank you very much, Jocelyn. Thank you, Paul.

Now, sadly, the pixel bots are a little bit too small for our keyboard players. So the University of Plymouth have donated part of their robot football team to the cause. And here they are. And these little fellows have just come back from playing football in China. And these six robots are able to share information between themselves. So when they get an instruction to play a note, say a key on a keyboard, the message is passed to all six robots.

And the closest robot to that key will play it. Now eagle eye, "Doctor Who" fans will have noticed that they've all come dressed as different incarnations of the doctor himself. And I think this one's my favourite, with a little scarf here. But let's hear them in action.

[KEYBOARD PLAYING]

That was pretty good. That was slightly imperfect, but not bad for some Doctor Whos, I think. Now, I think that's pretty good. Because our orchestra is almost complete. Towards the start of the lecture, I mentioned robots will be all controlled using MIDI information sent to each robot in real time that would tell them what to play. Well, once we'd gathered this many robots, we felt the time was right to plug them all together and try it out.

So what we should be able to hear is what we heard two days ago at the first rehearsal of our robot orchestra. And here is some footage we have. We connected everything together. And do we want to hear it?

Yes.

Yep? I warn you. It's not pretty. Here it goes.

[MUSIC PLAYING]

What do we think? Now, I'm not sure about you, but it didn't quite sound like the "Doctor Who" theme tune I've got in my head. And it's sort of in there somewhere, I think. But it wasn't great. So what was going wrong? And although we were telling the robots to play their notes at exactly the same time, some robots were taking slightly longer to play their notes, while others were doing it very quickly.

Now this gap is called latency, and it makes the orchestra sound out of time. So we had to go around to every robot and calculate how long to the nearest millisecond it took each robot to play a note. We then had to work out those figures into the score, meaning we'd queue some robots to play a fraction of a second before the others, so that when we hear the finished thing, they all sound like they're in time.

Now last, but definitely not least, when designing a robot orchestra, we wanted to include a percussionist that would be unashamedly cool and take to the air. Now you may have seen quadcopters for sale in toy shops, but nothing quite like this. Let's just watch what it can do.

The clever bit is nobody's controlling it. Instead, the flight of the quadcopter is being precisely controlled by a computer. And the software designed has been designed by a team at Bristol Robotics Laboratory. Now if you look above your head, some of you, you'll see some of these glowing red circles. And there's a few of them around the lecture theatre.

Now in fact, these are small infrared cameras, and they're looking for small balls just like this one, which have been placed around the quadcopter. These are motion capture reflectors. So the same thing that filmmakers would use to turn an actor into a CGI character. But we're using to tell the computer exactly where the drone is in this lecture theatre.

Now on screen, you can see the position of the cameras and the position of drone. Now if the drone stays in this precise location, in 3D space, those four motors can speed up or slow down and move it back to where it's supposed to be. But it's not just the quadcopter we can track. We've attached some motion capture reflectors to this teapot. And if wave this around, we should see the cameras are tracking the position of this teapot, as well.

And we've programmed this computer to recognise the movements of my teapot and adjust the positions of the quadcopter accordingly. So I should be able to control this quadcopter with my teapot. How good is that? That's fantastic. Now let's see how good I am at landing this.

[LAUGHTER]

Not so good. Now for our robot orchestra, the quadcopter's tether has been connected to this arm down here. Which is positioned so that when the quadcopter jumps into the air, it will strike and crush the symbol like so. So the flight path for each of these, for each of these jumps, has been preprogrammed and will be cued by a MIDI signal, just like all of the other robots. So it's completely automated. Right. I think our orchestra is just about complete. But we have a few more robot musicians I'd like you to meet.

So can we bring on the rest of the robots, please?

Now as well as Mortimer the drummer, we've also brought in this robotic drum kit that's been designed by a team at King's College London. So can we hear a little bit from this robot, please? Excellent. And at the back, next to the quadcopter, is a pipe organ that, believe it or not, used to be made out of a set of shelves. And it was hacked out of household items by a researcher at the University of Aberystwyth. And the air is actually being pushed through the pipes by an old vacuum cleaner.

So can I hear a tune out of the pipes, please?

[PIPES PLAYING]

Now that's how to hack your home. Very cool, indeed. We've also got a robotic glockenspiel and this very cool electric guitar, which uses compressed air to push down the levers on the strings. So can we hear the guitar?

[GUITAR PLAYING]

Excellent. OK. Now as promised, we've invited some human musicians to join our robots, too. So please can you welcome Galya, Robert, Chris and Kate, from the London Contemporary Orchestra.

[APPLAUSE]

Now I think we're actually there. So I'll just let you guys get comfortable there. And please welcome Andy Lambert from City University, London, who has been conducting this robot orchestra in the past few days. A huge round of applause for Andy, please. Now I've not actually heard these robots play together yet. So I genuinely have no idea how this is going to sound. So I'm with you guys tonight.

And even if the human outshine the robots, we have shown that performing in an orchestra is not beyond the grasp of machines. If we can master the technologies we've seen over the past hour, we should be able to make a better future for everyone, be that 3D printing limbs or discovering other planets. I call this lecture the new revolution. And the hacking revolution starts right here. I want you to stop thinking about your phone or your laptop as a black box, but something you can tinker with.

If your phone doesn't do what you want it to do, write an app of your own. If something in your house doesn't behave the way you want it to, think about how you can hack it, how you can take control. We set ourselves three grand challenges in these lectures, and my final challenge is to you. What is your great engineering challenge? What problem are you going to solve? And if you haven't worked it out yet, that's OK, too. Now is the time to get those skills, learn some code, buy that simple electronics kit. Then, when the inspiration hits, you will be ready, and sparks will fly.

Now I'm going to leave you tonight with the first ever performance of the Royal Institution robot orchestra. Are we ready for this? Yeah?

Yes.

Brilliant. OK, take it away, Andy.

[ORCHESTRA PLAYING]

[APPLAUSE]

Wow. Than you, and goodnight.

[MUSIC PLAYING]

Collections containing this video: