Like most people I’ve grown up with images of robots like this one from Lost in Space, but real robots are not like this. The real robots and the ones we are going to see in the next few minutes have got a great diversity of form and function. We are going to see robots that can move through corridors, we are going to see robots that can swim under water, robots that can fly, robots that can do assembly work. They all look very different and they are nothing like the robots that we have grown up with on TV and in the movies.
What’s common to all these robots although they are different, the common feature is that they perform fundamentally three actions. They sense the world and we are going to look at a variety of sensors that robots use to sense the world and then they make plan, they take what they perceive about the world through their senses and they have got a goal that they are trying to achieve. So they plan on how to move a bit closer towards the goal. And then they perform an action it might be to swim forward, it might be to fly, it might be to move down the corridor. And then it repeats; it senses again how the world is trying to adopt that plan and move some more. So robots are continually doing sensing planning and acting in a continuous loop. So lets go and look at some new robots.
This is an example of an industrial robot. Machines like this are direct descendants of the first robots that went into service in the 1950's. This is a very big industrial robot. Robots like these you will typically find in a car factory doing jobs like spray painting, welding, or lifting parts of car bodies. This particular robot can lift up to 60kgs. It's driven by 6 electric motors and they rotate the machine around different axes. So, the first motor rotates the robot about an axis like this; the next motor rotates it about an axis like this; next motor rotates it about an axis like this. Down here we have what we call the wrist mechanism. This assembly here; and that contains three motors and they rotate the end effector of the robot; the bit that does the work; they rotate it about an axis like this, like this and another one like this. So these six motors allow the end effector of the robot to move independently in the X direction, the Y direction and the Z direction. And also to rotate independently about the X axis like that; about the Y axis which is motion like that; and about the Z axis which is motion like that.
This is the Guiabot. This is a mobile robot. Unlike the last robot that we looked at this robot has the ability to autonomously move through an environment. To enable it to do that, it has a rich collection of sensors. This sensor is probably familiar to you, it’s a Kinect camera. We call it a 3D camera. It outputs the three dimensional structure of the world around the robot. And it needs that three dimensional structure so that it can determine what's an obstacle - a place that it shouldn't drive - and what's free space - that space where it is okay for it to drive. The robot also has got this pair of cameras here, we call this a stereo camera pair, and it mimics the function of our own two eyes. We use our two eyes to determine the 3D structure of the world. So the robot uses this to provide augmented 3D information about its world. The robots also got a very interesting camera on the top. We call this an omnidirectional camera. It's actually made up of six individual cameras which look in all these different directions as well as vertically. It integrates that together to give almost a hemispherical view of the world. A very, very wide angle view.
Other features of this robot are also interesting. Down here we have a sensor that we call a laser scanner. This is essentially an optical bumper bar. It is the sensor of last resort to stop the robot bumping into something that it shouldn't bump into. The robot has also got a lot of computing power onboard. We see here we have got two Mac Minis in addition to a computer which is in the base and that provides the computational power to process the information from all the sensors to plan how it should move and to control the motion of the wheels. This robot has got a very strong sense of where it is; it actually builds up a map of its world; a two dimensional and a 3 dimensional map. It knows where it is within that world and the fundamental information it needs in order to be able to achieve a goal. So it can sense, it's got planning computers here and that sends actions through to the motors and it continues to repeat these sensing, planning, acting paradigm.
Here we have another mobile robot and this is a giant one. Machines like these are called hot metal carriers. They are a relative of the forklift truck. They are typically found in aluminium smelting facilities and their job is to pick up very large buckets like this one you can see in the foreground here which will be filled with molten aluminium - 8 tonnes of molten aluminium. Their job is to pick up this bucket, lift it and carry it along a road network to another place within this smelting complex.
This machine was originally driven by a person who would sit in a cab here and manipulate the controls, but this particular machine has been turned into a robot. In order for if to be an effective robot, it needs to use a number of different sensors so that it avoids hitting things or people that it shouldn't hit, that it can determine where it is within the smelter complex, and also to be able to successfully pick up the crucible.
Now, let's look at the pick up process. The vehicle itself has got this hook which it needs to put in through this handle. So it's a relatively precise positioning problem that the robot needs to solve. Now it does that by using a camera which is up here on the mast of the robot, which looks down and works out where this bucket is with respect to the robot, and the robot moves so as to bring this hook in through this handle.
Another really important sensor on this robot is this yellow laser scanning device, and there’s actually four of them, one on each corner of the robot so that it can see all the way around. This performs two functions. One is the laser beams can detect something like me, an obstacle, and it would then stop so it wouldn't collide with me. But it also performs a really important navigation task. This is how the robot knows where it is within the complex. So to understand how that works we are going to go outside and look at one of the navigation markers that this sensor picks up.
The robot works out where it is in the environment using a system of markers. In this particular case these strips of reflective tape which give really bright returns on the laser scanners those yellow devices that we saw on the corners of the hot metal carrier. Now the robot has got a map that tells it whereabouts all these markers are within the complex. And then it uses the information about where the robot perceives these markers to be and it combines both of those together to determine where the robot is with respect to the map.
Here we have another kind of mobile robot. This one is mobile in the air. It's a small scale helicopter it's a sort of thing that a hobbyist would own. Sort of thing that they might fly on the weekend and they fly it using joystick controls that requires quite a lot of skill on their part in order to keep it in the air and go where they want it to go. This particular machine has been converted into a robot but before we talk about the robotic part, just talk a little bit about the helicopter machine itself because there are fascinating things.
This is powered by a small petrol engine; there is a fuel tank down here that carries the fuel. And the main lift force comes from the main rotor that provides a force vector vertically upwards, which counteracts the weight of the vehicle - counteracts gravity. The amazing thing with the helicopter is it can point the force vector in different directions. It can point it forward, it can point it backwards it can point it sideways and that's all done by this mechanism here called the swash plate and what does is every time the blades rotate, they also rotate this way about their axis so that allows this force vector to be controlled and that gives it the ability to move forward, backwards, sideways and hover. There is another propellor on the back here and that provides a force to stop the whole body of the helicopter rotating around.
In order to turn it into a robot, we need to add some sensors to it. So in this box down here, there are some sensors such as GPS which tells it where it is with respect to the world. There are some accelerometers and gyroscopes that tell it it's orientation in space. There is a laser altimeter here which tells it it's height above the ground and in the front is a radar which detects obstacles that it might fly into. A robotic helicopter like this is useful for many kinds of tasks. It's useful for automated inspection: perhaps on a farm it could inspect the quality of crops, it could fly along creeks and waterways, it could fly along electricity lines or along fence lines.
Here we have another type of mobile robot but this robot is mobile through water. We use robots like this to make scientific measurements about large volumes of water like lakes or coastal zones. It looks quite simple but exactly quite complex and sophisticated machine. Lets walk through it. Down at this end we have a propeller that pushes the whole robot through the water. It's got some control fins, rudder is here to steer it through the water body and these planes here like it move up and down in the water column. This antenna stack here it's got a GPS receiver so when it's on the water surface it knows where it is, and it has got some other antennas that allow us to communicate with it when it is out at sea. Here we have sonar sensor - side scan sonar - which uses sound waves to build up a three dimensional map of the sea bed as the robot passes over it. Up here we have some sensors that measure the speed of the ocean current. Even though the robot is moving that can measure the speed of the water column. Underneath there is another sensor that measures the speed of the robot with respect to the water column. Up the front here is the scientific payload and this is the instrumentation that we want to push through the water to make measurements about what's going on. So here we have sensors that measure the conductivity and pH of the water; that measure the amount of dissolved oxygen; that measure the amount of chlorophyll; measures the amount of blue green algae. So this scientific data is merged with navigational data which comes from the other sensors on board the robot, so we know where these measurements have been taken. This particular robot has got a runtime of up to 8 hours, moves about 2 meters per second, so we can survey about 30 kilometres in water in a single run.
So what we have just seen are a number of really interesting, exciting and clearly useful robots. These robots differ in their form and in their function, but inside these robots are all doing the same thing. They are sensing their world, they are making plans and then they are they are carrying out actions in order to achieve their goals. Some of the robots that we saw were clearly research prototypes, but others were commercial products. And this is really exciting because this is the beginning of a whole new technological wave and we have witnessed many technological waves in the past. The introduction of steam power, electricity, flying machines, automobiles, electricity, computers; all these things were surprising to our grandparents and to our parents. Robots are just going to be the next technological wave and in the near future, we are going to be quite unsurprised to see them in our workplaces and in our homes. So there are clearly really interesting times ahead.
In this mini-documentary we look at a diverse range of real-world robots, discuss what they do and how they do it.