A Dynamics and Task Decoupled Reinforcement Learning Architecture for High-Efficiency Dynamic Target Intercept

Due to the flexibility and ease of control, unmanned aerial vehicles (UAVs) have been increasingly used in various scenarios applications recent years. Training UAVs with reinforcement learning (RL) for a specific task is often expensive terms time computation. However, it known that main effort process made fit low-level physical dynamics systems instead high-level itself. In this paper, we study apply dynamic target intercept (DTI) task, where equipped by different UAV models are correspondingly distinct. To end, propose decoupled RL architecture address inefficient procedure, module focuses on modeling DTI without involving dynamics, design states, actions, rewards completely task-oriented while control can adaptively convert actions from signals retraining module. We show efficiency efficacy our results comparison ablation experiments against state-of-the-art methods.