Narrator:
This is Valkyrie, a humanoid robot made by NASA. Its metal bulk means that, even at 170 cm tall, it weighs four times as much as the average 11 year old, weighing in at 120 kg. It's the result of a collaboration between NASA and the University of Edinburgh on a very ambitious space project. NASA will send a team of Valkyrie robots to Mars to prepare things on the hostile planet's surface ready for the arrival of their human colleagues. The reason it looks so human is so it can use exactly the same equipment as the astronauts it will be working with. Valkyrie has 44 movable joints, giving it very similar levels of dexterity to a human being. The team working on Valkyrie have to develop new techniques and algorithms to help the robot's walking, manipulation and balance control, all things that we take for granted as human beings. The Valkyrie is at the forefront of a new frontier for robotics working towards a future where robots and humans work in collaboration, side by side, to achieve great advancements in space exploration.
Dan Walker:
That music is just magnificent isn't it? We're joined now by Professor Sethu Vijayakumar: and some members of his amazing team from Edinburgh University. Valkyrie. I mean what a bit of kit that is.
Sethu Vijayakumar:
Absolutely. I just got an email yesterday from Kim Hambuchen, the Director of the Space Voyage Challenge saying, "Your girl's flown."
Dan Walker:
Okay. So I need to process this. You got an email from NASA. And when they say, flown, how far through the process? There's still lots of work to do isn't there?
Sethu Vijayakumar:
Yeah. So actually it's not flown to Mars. So actually it's on its way to Edinburgh and, over the next four to five years, we are going to work, and the team is going to work, on algorithms for making the robot walk, manipulate, do the things that you and me do but do it much more efficiently and, as a roboticist, I'm really excited about it.
Dan Walker:
You've got a robot nut here as well. [LAUGHS]
Ana Matronic:
I'm really excited about that. So, in the next four or five years, I'm gonna come to Edinburgh and, hopefully, touch your Valkyrie. Now I love all robots. Valkyrie is state of the art but some of her processes can be made quite simply with stuff at home right?
Sethu Vijayakumar:
Absolutely. So robotics, you can make it as complicated as a Valkyrie or as simple as something here so you can really suit the projects that you do in robotics to your budget and your talent.
Ana Matronic:
And now I see, on the base of these things, those look like toothbrushes to me. This looks like something you really can make at home.
Sethu Vijayakumar:
That's correct. So it's made out of two toothbrushes, a battery pack and your microbit. So if you make this, it's basically an off centred motor. If it's wired right it can basically do a little jig of its own.
Ana Matronic:
Fantastic. Alright, now don't go getting all your family's toothbrushes and cutting them off. We have instructions on our website. Work sheets show you how to create these little micro-bots and they're all on our Live Lessons website.
Dan Walker:
What are these funky little guys? What are they doing?
Sethu Vijayakumar:
So actually Dan, these are commercial versions of what we just built. They're called the Hexbug and the principles are again exactly the same. You have a little motor that vibrates and here its completely open loop means that it reacts with the environment and does what you think is a really fancy move but it is actually completely open loop.
Dan Walker:
Okay.
Ana Matronic:
So a very simple principle. But we've got something a little bit more advanced here.
Dan Walker:
Yeah the big boy.
Ana Matronic:
A humanoid robot.
Sethu Vijayakumar:
That's correct. So this is the next step up from a bug.
Ana Matronic:
Okay.
Sethu Vijayakumar:
So this, as you can see, is a mini humanoid called Condo and it's got 19 joints of, what we call, the degrees of freedom and, actually, you can basically program it with very simple principles, exactly as the robot there. You can take each of the joints and the difficult bit is to sequence all of them to actually do a little complicated job.
Dan Walker:
Yeah, let's see what he's got.
Ana Matronic:
Oh wow. What a gentleman.
Sethu Vijayakumar:
So I think, as you can see here, you can get it to do rather complicated things but, what's slightly complicated and more difficult to explain is that, while he's doing this, if I be a bit nasty and give it a push, it doesn't recover from it so it can only repeat the same things again and again and really do it very well but not react to external forces.
Ana Matronic:
Okay.
Dan Walker:
I absolutely love that. You've got to be careful. She will take that home with you right? [LAUGHS]
Ana Matronic:
Yeah. You might have to search my bag.
Dan Walker:
We're gonna let you go 'cause you've got something funky to put on for us in a minute.
Ana Matronic:
I do.
Dan Walker:
We'll be back with Ana in a moment. And Sethu, if we come down here we've got a sort of, I suppose, the next level up. That's right isn't it? And, Sandy one of your PHD students, who's not a robot, he's a human, I have checked, is here as well. And talk to us a bit more about this. This is like the next stage in, I suppose, evolution of robots you could say.
Sethu Vijayakumar:
That's correct. So, so far so for all the robots that we've seen earlier, they were pretty much repeating what you've got to do without any human input. So the next couple of robots you're gonna see are things where humans, so for example you and me could change the behaviour of the robot by an external input, and I think the first thing we're gonna do is to get Sandy to explain a little bit about Archie, the robot and he's a developer of that.
Dan Walker:
'Cause you've got a microbit in your hand there haven't you?
Sandy Enoch:
I do yeah. So I'm using the microbit here to actually control the robot, so as a control for it. So we're sensing the tilt of the microbit and using that to steer the robot around. So this is kind of like the sensor you could use on the robot to sense how it was balancing and we can add that information in and get the robot to actually listen to its environment as well as do things like play games like football.
Dan Walker:
Okay. You're slightly blowing my mind. What I find annoying about you and impressive at the same time, is that you did this in your spare time. This wasn't in your PHD. [LAUGHS]
Sandy Enoch:
Yes, this was a side project I did but it was more fun than my PHD so, you know.
Dan Walker:
You've got some good students, yeah.
Sethu Vijayakumar:
And actually this bit uses the accelerometer from the microbit but we can also use the magnetic sensor and I could build another piece of equipment which is basically using this like a magic wand to pull and push this kit. So it's really the on board sensors can be used in very, very creative ways to control and create input to the robot.
Dan Walker:
Okay. Sandy you keep playing with Archie. Let's move Sethu along the chain. Now this is a game which people recognise. We've got Eleanor here who's playing a robot at this game which many people have played at home. This looks like the next level up. What's happening here?
Sethu Vijayakumar:
So this is a Connect 4 robot obviously but, basically, this really captures the essence of all of robotics. So there is sensing which is a camera looking at the state of the board and making basically the sense of the state of the [UNSURE OF WORD] of the board. You've got a robot arm which is moving and picking up pieces and playing the game and there's an AI which is connecting the sensing to the actuation to figure out what is the next best move to make.
Dan Walker:
Okay. Let's find out from Eleanor whether she's managed to beat the robot yet. Any chance?
Eleanor:
No.
Dan Walker:
Has it wiped the floor with you so far?
Eleanor:
Yeah.
Dan Walker:
Okay. Is that because it's unbeatable Sethu? Can Eleanor beat it? Can somebody beat it?
Sethu Vijayakumar:
So Connect 4 is what is called a solved, so completely, strongly solved game which means that, in theory, if you start the game as if you are the first move, then you can win but, although Eleanor has started, if you make a single mistake, the robot will get you.
Dan Walker:
Okay, Eleanor, I'm sorry. You've basically got no change whatsoever. Although there is a flaw in your plan. You haven't named this robot yet?
Sethu Vijayakumar:
That's correct. How about we suggest something?
Dan Walker:
This is a big moment. Roger the Robot. Is that right?
Sethu Vijayakumar:
That's a very good name.
Dan Walker:
Okay excellent. I thought we might get a round of applause for Roger the Robot. Is that-- no? Thank you. Thank you. Thank you everyone. Again, one of the essential things about this is these are principles here which are the same, which you've used in the microbit, which go right the way up to Valkyrie don't they?
Sethu Vijayakumar:
Absolutely. So this really embodies the fact that, when you find something that's unexpected, the robot has to react and change its behaviour and, for example, if you send a robot to Mars, you cannot possibly know all eventualities and you cannot program for all these things. And this is an example of how a robot reacts to how a human plays and changes its behaviour based on that.
Dan Walker:
The good thing is though, Eleanor was smart enough to notice that Roger the Robot had run out of counters and you put the counters in so, without the human element, it wouldn't have worked. We still need humans. So Sethu now, again, this is where we're gonna bring Ana back in because something funky is happening over here. You need to explain and talk us through exactly what this is.
Sethu Vijayakumar:
So we've actually managed to finally build a robot hand that matches Ana's passion, her fame and her tattoo.
Ana Matronic:
Yes.
Sethu Vijayakumar:
So we've got this prosthetic hand which is actually things that, for example if you have lost your limb in an accident or if you're born with a congenital defect, then you can actually use this to restore the upper limb, sorry the motor capabilities, of just using this prosthetic hand. So what I'm gonna try and do is to try and control Ana's prosthetic hand鈥
Dan Walker:
With your arm, yeah?
Sethu Vijayakumar:
鈥ith my arm. So there are these END sensors which are sensing the muscle activity on my hand and really the hand is responding to that. So we're going to try and attempt to lift this ball up and hand it over to Dan.
Dan Walker:
Thank you very much.
Ana Matronic:
Oh, that's so cool.
Sethu Vijayakumar:
And we're gonna try and attempt to pick up this bottle and, again, hand it to Dan.
Dan Walker:
Thank you robot arm.
Sethu Vijayakumar:
So this is an example of what we call shared autonomy where the intentions of the human motion is transferred to the smarts of the hand and a lot of the elements, in terms of the control, is happening in the hands.
Ana Matronic:
And now you have a student with us today. This is Iris who is working on a more advanced version of this technology so tell me about this.
Sethu Vijayakumar:
That's correct. So Iris is a [UNSURE OF WORD] student who is working on the next generation. So, in the previous example you saw, that I had to actually move and send the open and close signals. So, here, the system that Iris is working on is pretty much actually detecting the intention of the person wearing the prosthetic to figure out whether he's going to lift a bottle or a CD case or a credit card and change the grip and pre-shape the grip so that the person who's using it can use it much more effectively and much more easily.
Ana Matronic:
Wow. So these sensors that I see on her arm are obviously much more sensitive than the one that you have and they're sensing the micro transmissions of her muscles.
Sethu Vijayakumar:
Absolutely.
Ana Matronic:
That's amazing. It's incredible. It look like you're moving it with your mind. [LAUGHS] it's fabulous. Wow. It's amazing.
Narrator:
Did you know the first working robot was used to make cars in a Ford factory in 1961? Today there are over a million robots in use across the world, with over half being used in Japan. The smallest robot is called a nanobot. It measures 10 nanometres in size which is less than one thousandth of a millimetre.