2P1-F21 物理モデリングによる飛行船の追跡行動の獲得

概要

論文の詳細を見る
We have proposed an approximate fluid environment by using the physical engine, PhysX. The physical engine enables the airship model to simulate its actions in fast, real-time, and dynamic simulation. This study aims at an acquisition of autonomous behavior for the airship to pursue the target under the virtual air environment. The autonomous airship actions are obtained by a controller ANN(Artificial Neural Network), which is evolved to learn proper actions by GA(Genetic Algorithm). Simulation results shows that the airship acquires appropriate actions under the virtual air environment.