Hierarchical reinforcement learning guidance with threat avoidance

The guidance strategy is an extremely critical factor in determining the striking effect of missile operation. A novel law presented by exploiting deep reinforcement learning (DRL) with hierarchical deterministic policy gradient (DDPG) algorithm. reward functions are constructed to minimize line-of-sight (LOS) angle rate and avoid threat caused opposed obstacles. To attenuate chattering acceleration, a structure improved function action penalty put forward. simulation results validate that under proposed method can hit target successfully keep away from threatened areas effectively.