Responding to a changing environment

When you send a robot out into the real world, as it were, you’ve got certain things that you know about that are fixed. Like trees don’t tend to move and buildings don’t tend to move and things like that. But there’s lots of things in the real world that do move, predominantly things like people, and vehicles, and animals, and those sorts of things. And what you really don’t want is, you don’t want a robot that just forgets about those things and just crashes into them, or drives into people, or anything like that. Because that’s not really a very good robot. It’s not very good at its job.

So it’s those dynamic parts of the environment that are really challenging for robots, because it’s not only about knowing where they are, it’s knowing about what they’re going to do next and where they’re going to go, so that you can behave almost socially, so the robot can interact socially with these elements in the environment. What we’ve been doing with the youBots is - they have a sensor on the front of them, which is a laser scanner sensor, so that can detect other objects in the environment.

Now, you can simply be very reactive and say, “I’ve seen something, therefore I should stop,” for example, but that’s not very useful, because that means the robot doesn’t really get on with its job then. So, what you want to do with these is you want to say, “I’ve seen something, I think it’s a person, and I think this person is going to walk in front of me, so I’m going to stop and go behind them, or maybe I can go a bit faster and go in front of them,” and just interact in a very natural way, just like humans do. When we see another person, we just don’t panic and stop. We re-evaluate the situation and walk around.

And that’s the sort of work we’re trying to do with the youBots. We have a robot that we’re trying to send out as a guide robot. Our lab here in Sheffield is actually quite difficult to find. So, we want to be able to send a robot to the front of the building to collect visitors, so we don’t have to keep going out there and doing that. But in order to do that, it has to go through a number of buildings that have got students in, they’ve got other staff in, there are even other departments who don’t know about the robotics that are going on. So it’s a very challenging environment for that robot to have to navigate through.

The robot that we’ve been using is called the KUKA youBot. It’s a small-scale version of a large industrial robot. They use it in warehouses and factories - the larger version - to move things around. At the moment, these systems are remotely operated. But what we are trying to develop, as we’ve talked about today, is more autonomous manoeuvring for them, so it can free up the human operators to do other tasks rather than just driving the robot around for them. One of the big benefits with these robots is that they can actually move in all directions. So it’s very useful in a manufacturing setting, being able to drive sideways.

They have some very clever wheels that allow them to drive in all directions. One of the reasons that that’s very useful for our work, as well, is that it allows the youBot to manoeuvre very similar to a drone. Obviously only in two dimensions. It can’t go up and down like a drone, but it can manoeuvre sideways and backwards and forwards and everything, just like a drone.

And a lot of the technologies that we’re developing in terms of working in dynamic situations - ultimately, we want to apply them to drones, to flying systems - to be able to avoid other aircraft, or birds, or things that might appear in the environment when you’re trying to do, for example, a building inspection with a drone, or something like that.