Design of a Deep Reinforcement Learning Algorithm for Spatially Adaptive UAV Autonomous Navigation based on Transfer Learning

Puntos clave

• The research aims to enhance exploration efficiency in UAV autonomous flight using a deep reinforcement learning algorithm with transfer learning.
• Developed a deep reinforcement learning algorithm combining DQNs and transfer learning.
• Tested the algorithm in diverse dynamic environments.
• Evaluated the efficiency of exploration and adaptation capabilities.
• The proposed algorithm demonstrated faster convergence than traditional reinforcement learning methods.
• Achieved superior exploration performance in various operational areas.
• Showed practical applicability for military and civilian UAV missions.