So let’s look at feedback control in a bit more detail, going back to the steersman. So we’ll zoom in to the top image. Because if the steersman is going on the right course, that’s fine. But if he notices that he’s starting to the left, for instance, he adjusts the rudder to steer back so he’s back on course. And if the wind suddenly veers up so he goes in the opposite direction, he adjusts the rudder in the opposite direction. And the diagram shows this. We can see the boat at the top, which is affected by the winds and tides. And the output is the course.
And if, for instance, the boat goes one way, the plus indicates that, the steersman adjusts the rudder in the opposite direction and you see the minus. The net result is you have a feedback control. Now let’s apply that idea of feedback control to a manipulator robot. So we’ve got an arm– and I’m emulating a robot here– with a gripper at the end. And you want the gripper to be in the right place, avoiding any obstacles. How is that achieved? Well, the robot determines where the gripper is. Is it in the right place? No, it’s not. It adjusts the motors until the gripper is correctly positioned. Same idea, same block diagram applied, but it’s a bit more complicated.
Because for the gripper to be in the right place, the wrist, elbow, and shoulder joint each have to be at the right angle. So we need feedback control applied to each of those joints. But how does the robot determine how it moves in its environment? Should we pre-program it? Or would it be more interesting to see if the robot can learn its behaviour? Now, learning, you’ll be pleased to note, is a feedback process. Which we can again illustrate with the same form of block diagram. Because if you do something, you then think about it. Well, how well did I do that? You are evaluating what you’ve done. You’re measuring your performance, if you like.
And on the basis of that, you adjust your behaviour to do better. Learning by your mistakes or by trial and error. So feedback is also used for learning. What allows us to learn? Well, we’ve got a very sophisticated brain. But for a robot, we haven’t got the ability to have that most sophisticated system. So instead, we can have very simple brains. We’ll see how, with just three or four so-called neurons, you can get a robot to move around, exploring its environment, using lights or part of cybernetics and artificial intelligence, fitting in with robots. The other use of feedback is interaction. The robot moving around is interacting with its environment. The steersman is interacting with his boat.
So we’re interested, very much, in how machines interact with humans or with their environment. The ultimate interaction between a human and a machine is a virtual reality system. There, for instance, the computer generates an image of an artificial world which you display to the human. And if the human turns their head, the world looks differently. Therefore, the computer needs to know that the human has moved. And therefore, can respond accordingly. So therefore, my block diagram, the computer and the movement, it’s still the same basic idea about feedback systems. So in this course, we’ll show some Interaction. We’ll have a number of our robots interacting with each other. We will see robots interacting with humans.
We will use, for instance, our Baxter robot, which is specifically designed to work with humans. And then we’ll consider another form of interaction, using Haptics, which also comes from a Greek word. In normal virtual reality systems, you can see what it looks like. You can perhaps hear what it sounds like. But how does it feel? Well, we can use devices, which look a bit like robots, which you grip. And they can move subtly, which will give you the impression of feeling something that isn’t , in fact, there. And finally, we consider sensors and electronics. If our robot is to move around its environment, it needs to be able to see what’s there, to detect how far away objects are.
It uses sensors for this purpose. These produce little signals which we then need to process using electronics to make them bigger, to amplify them. And again, we use feedback to amplify signals. So in this course, we’re going to consider robots, primarily mobile ones. What is in a robot? What is their anatomy? The mechanics, the electronics, the sensors, and so forth. How do we ensure that they drive at the right speed? Using feedback, of course. And what is their behaviour? How can we decide what the robot does? How can we get the robot to work out for itself what it should do?
And this is going to be achieved by a few talks, some interactive web-pages, some screencasts, some articles for you to read, some external websites for you to visit, and some quizzes for you to do. We hope you enjoy it.