Dutch air force reads pilots’ brainwaves to make training harder

Fighter pilots in training have their brainwaves read by AI when they fly in virtual reality to measure the difficulty they find in tasks and increase the complexity if necessary. Experiments show that fighter pilots in training prefer this adaptive system to a rigid, pre-programmed alternative, but that it does not necessarily improve their skills.

Pilot training in simulators and virtual reality is cheaper and safer than real flights, but these teaching scenarios must be adjusted in real time so that tasks fall between comfort and overload.

Evy van Weelden of the Royal Dutch Aerospace Center in Amsterdam and her colleagues used a brain-computer interface to read student pilots’ brain waves via electrodes attached to the scalp. An AI model analyzed this data to determine how difficult it was for pilots to find the task.

“We are continually working to improve [pilot] training, and what that looks like can be very different,” says van Weelden. “If you’re not in the field, it’s very science fiction-like, I suppose. But for me it’s really normal because I only see data.

Fifteen Royal Dutch Air Force pilots underwent training while the system toggled between five different difficulty levels – increasing or decreasing visibility in the simulation – depending on how difficult the AI ​​model was determining their missions.

In subsequent interviews, none of the pilots said they noticed the system changing the difficulty in real time, but 10 of the 15 pilots said they preferred the changing tests to a pre-programmed exercise where the difficulty gradually increased in regular steps.

Importantly, none of the pilots showed any improvement in task completion under the adaptive simulation compared to a rigid simulation. In short, pilots liked the mind reading setup, but it didn’t make them better pilots.

The problem could lie in the unique nature of individuals’ brains, says van Weelden. The AI ​​model was trained on data from another group of novice pilots and then tested on all 15 study participants. But it is notoriously difficult to run AI models that analyze brain waves on the entire population. Six of the pilots participating in the test showed little change in difficulty level readings, indicating that the AI ​​system may not have interpreted their brain data correctly.

James Blundell of Cranfield University, UK, says similar technology is being explored for use in real planes to ensure pilots remain in control. “They looked at whether we could detect a startle – like a bit of panic – and what the plane could then do to calm you down and then reorient you,” Blundell says. “So you’re upside down, [and the aircraft might say] you really have to look at the attitudes, you have to look at the information that’s here, that’s going to bring you back to the right path.

These systems have shown promise in isolated scenarios, but it remains to be seen whether brain reading technology can be used to improve aircraft safety. “There is a long way to go [in order to achieve that]” says Blundell.