The rearing of aircraft pilots requires a lot of cost and time, so it is highly important to improve the training efficiency. To efficiently improve pilot abilities, appropriate instruction by experts is necessary, but the shortage of such experts is becoming an issue due to the aging of pilots. Therefore, for the purpose of training, we have developed an inverse reinforcement learning method to learn piloting know-how based on information from instruments and visual information that experts pay attention to during a piloting operation. We applied this technology to aircraft landing problems and confirmed that it enables learning of experts' landing skills. This report also describes a pilot training system that utilizes the obtained piloting models.
