Considered much safer than nuclear fission, fusing hydrogen atoms offers a potential route for scientists to achieve a source of perpetual, renewable energy.

The only problem is, fusing hydrogen atoms is very complex.

There’s hope that the deployment of Google-backed AI, DeepMind, to the tokamak (magnet-based) nuclear fusion reactor, will help scientists gain a better understanding of how plasma interacts inside a reactor under various conditions.

DeepMind has been working with the Swiss Plasma Center (SPC) reactor at École Polytechnique Fédérale de Lausanne (EPFL). This reactor is known as a variable-condition tokamak (TCV) – differentiating itself from others by allowing for a wide range of plasma configuarations. However, actually utilising this properly is a huge – and hugely expensive – undertaking.

On this, SPC Scientist Federico Felici, said: “Our simulator is based on more than 20 years of research and is updated continuously.

He added: “But even so, lengthy calculations are still needed to determine the right value for each variable in the control system. That’s where our joint research project with DeepMind comes in.”

The Google AI was put through a variety of control scenarios within the SPC’s simulator. Through this, the reactor’s ML system was eventually able to calculate control strategies for producing desired plasma configurations.

Recommended

• Leader Insights | Demystifying tech to ease the skills shortage
• Open banking hits 5 million UK user milestone
• Visa and amazon have resolved their credit card payment dispute

From this, DeepMind was deployed in the actual tokamak reactor, where it was able to create and control a variety of plasma shapes.

Speaking about the progress made, research scientist at DeepMind, Martin Riedmiller, said: “While there is still much work to be done… we are pleased that the results indicate the power of AI to accelerate and assist fusion science, most likely augmenting human expertise in the field and serving as a tool to discover new and creative approaches for [fusion reactor control] and beyond.”

He added: “[Our findings] also suggests that there might be potential for wider adoption of deep reinforcement learning on physical systems for complex scientific and industrial machines, from simple motor control to complex robots.”

Get the latest news from DIGIT direct to your inbox

Our newsletter covers the latest technology and IT news from Scotland and beyond, as well as in-depth features and exclusive interviews with leading figures and rising stars.

To subscribe, click here.