Dutch air force reads pilots’ brainwaves to make training harder

Fighter pilots undergoing training are now having their brainwaves monitored by artificial intelligence (AI) while flying in virtual reality to assess the difficulty of tasks and adjust complexity levels as needed. Recent experiments have shown that trainee pilots prefer this adaptive system over a rigid, pre-programmed one, although it does not necessarily enhance their skills.

Training pilots in simulators and virtual reality environments is a cost-effective and secure alternative to actual flights. However, these training scenarios must be dynamically adjusted to ensure that tasks are challenging enough without overwhelming the pilots.

Evy van Weelden and her team at the Royal Netherlands Aerospace Centre in Amsterdam utilized a brain-computer interface to capture the brain activity of student pilots through scalp electrodes. An AI model then analyzed this data to gauge the level of difficulty faced by the pilots during the tasks.

According to van Weelden, “We are constantly striving to enhance pilot training, and the methods we employ can vary significantly. While it may sound futuristic to those outside the field, for me, it is a routine part of my work as I focus on interpreting data.”

In a study involving fifteen Royal Netherlands Air Force pilots, the adaptive system alternated between five difficulty levels by adjusting the visibility within the simulation based on the real-time analysis of the pilots’ brain activity.

Although none of the pilots were aware of the system’s dynamic adjustments during the training sessions, 10 out of the 15 pilots expressed a preference for the adaptive tests over a pre-programmed exercise with incremental difficulty increments. However, there was no significant improvement in the pilots’ task performance within the adaptive simulation compared to the rigid one. In essence, while pilots appreciated the brain-reading technology, it did not translate to enhanced piloting skills.

The challenges may stem from the unique characteristics of individuals’ brains, as noted by van Weelden. The AI model was trained on data from a different group of novice pilots and then tested on the fifteen participants in the study. It is notoriously difficult to develop AI models that can accurately interpret brainwave data across a diverse population, as evidenced by six pilots in the study showing minimal changes in difficulty level readings.

James Blundell from Cranfield University in the UK highlighted ongoing research on similar technologies for use in actual aircraft to ensure pilots remain in control. These systems aim to detect signs of distress or panic in pilots and provide corrective measures to stabilize the aircraft. However, while these systems have shown promise in specific scenarios, their effectiveness in enhancing aircraft safety remains uncertain.

Blundell stated, “There is still a long way to go before brain-reading technology can be effectively utilized to improve safety in airplanes.”