Path Planning of Unmanned Helicopter in Complex Dynamic Environment Based on State-Coded Deep Q-Network

Unmanned helicopters (UH) can avoid radar detection by flying at ultra-low altitudes; thus, they have been widely used in the battlefield. The flight safety of UH is seriously affected moving obstacles such as flocks birds low airspace. Therefore, an algorithm that plan a safe path to urgently needed. Due strong randomness movement bird flocks, existing planning algorithms are incompetent for this task. To solve problem, state-coded deep Q-network (SC-DQN) with symmetric properties proposed, which effectively randomly and UH. First, dynamic reward function designed give appropriate rewards real time, so improve sparse problem. Then, state-coding scheme uses binary Boolean expression encode environment state compress space. encoded input learning network, important improvement traditional algorithm. Experimental results show SC-DQN help unknown motion status safely successfully complete raid