Netherlands Air Force Tests EEG-Based VR Pilot Training
The Netherlands Air Force is testing VR pilot training using EEG brain-wave data and machine learning to automatically adjust difficulty, with mixed performance results.
The Netherlands Air Force is experimenting with a new pilot training system that analyzes pilots' brain activity and automatically adjusts task difficulty. The training takes place not in the sky, but in virtual reality — inside a simulator cockpit designed to closely replicate a real aircraft.
During a VR flight, the pilot wears an electroencephalograph (EEG), a device that records brain waves. The data is processed in real time by a machine-learning algorithm. Its purpose is to determine how difficult the current task is for the pilot. If the system detects that the pilot is handling the exercise too easily, the next task becomes more difficult. If the workload is too high, the difficulty level is reduced.
Difficulty in the simulator is changed through flight conditions. For example, visibility is degraded, fog appears, reference points disappear, or the horizon line is distorted. These factors create additional challenges and test spatial orientation skills, which are especially critical for military pilots.
Pilots from the Royal Netherlands Air Force took part in the experiment. They performed identical tasks in two modes: one with a pre-set difficulty sequence and another using an adaptive system that adjusted to their condition. Afterward, participants evaluated workload, training comfort, and their own performance.
The study found that automatic difficulty adjustment did not produce a noticeable improvement in flight skills compared to the conventional approach. However, the pilots generally preferred the adaptive system. They said this training format felt more logical and more personalized.
The project’s authors note that simulators and virtual reality make it possible to train pilots more cheaply and safely than real flights. However, effectiveness depends on accurately selecting the workload level. Using brain-activity data may become one way to make training more flexible, although the technology remains experimental and requires further research.