Piloted simulation evaluation of a neural network limit avoidance system for rotorcraft

Piloting a rotorcraft is typically a high gain task, and under adverse conditions the workload may increase to the extent that the pilot is not able to achieve all goals simultaneously. Increased exceedances of operating liits may occur as a result of this high workload. The goal of this research has been to implement in piloted simulations a system that reduces workload and helps pilots to avoid aircraft limits. The system uses neural networks to predict near-future limit exceedances, and alerts pilots to these impending exceedances through tactile cues on the control inceptors and visual cues on the head-up-display. The system was demonstrated in piloted simulations of the UH-6OA and OH-58D and was found to reduce limit exceedances and pilot workload. In these experiments, tactile cues alone generally performed better than visual cues alone, but the combination of visual and tactile cues generally performed best. Pilot comments and handling qualities ratings of the system were highly favorable.