The Not-Quite-Robots That Help Fix Fusion Reactors

I am not inside a fusion reactor. It'd be wonderful if I was, the Joint European Torus reactor here at Culham Centre for Fusion Energy is behind some blast doors just a few metres away, but that's not really somewhere humans are meant to go. Not even for maintenance: it's not really dangerous in the same way that current nuclear power plants are, but it contains toxic beryllium and low-level radioactive… stuff. This is the mockup version of the reactor, used for training. And by the way, if the camerawork here seems a little bit odd, that's because it's being done by… well, here. Meet my camera operator. Say hi! I was going to say that my camera operator is a robot, but that's not strictly true either. This isn't a robot, it can't carry out anything automatically: instead, it's a remote handling arm. Basically, it's an extension of John over in the control room. It's not as versatile as a human arm, but it's got enough degrees of freedom.

You are actually there, at that point, and not detached from it in a control room. You don't even realise you're doing it. You're actually thinking about the job as you're there, and you're imagining all the components around you. Certainly from the views, we have a VR model as well, which gives us good detail, so from that you actually work at the coalface imagining that the items are there right next to you. It's got very good feedback. Even down to something as thick as a layer of tape. If you were to put that on a smooth surface and run the grip across it you'd be able to feel that level of detail. We have to service the ceiling as well, so the transporter will drive the manipulator upside down and up into position so the operator stays in this position, but gravity changes. So we have to teach the operators to deal with that as well. There are all sorts of cool technologies for remote presence that could be used one day.

Virtual reality. Gesture recognition. 3D cameras. All of which would be great: if they work reliably, every time, and if they'll still work and be supported after 10, 20 or 30 years of use in a radioactive environment. This system can't be experimental, it can't be risky, because the only way to fix or upgrade this robot arm inside a reactor is to send in a human. You don't take risks for something like this: you use something that is known to work safely. And I did ask whether I could have a go controlling the remote handling arm. For all those same reasons, the answer was yes… but very, very slowly. [Translating these subtitles? Add your name here!].