Trajectory tracking control of an unmanned aerial vehicle with deep reinforcement learning for tasks inside the EAST

•We propose a deep reinforcement learning-based UAV system that can conduct trajectory tracking tasks along a set trajectory in the vacuum vessel of the EAST.•To simulate the experimental environment, we built a gym-like simulation environment using Pybullet as the physics engine for simulating the EAST vacuum vessel model and UAV model. In this environment, the UAV dynamics model and the task and reinforcement learning network architecture are defined.•To verify the transferability of the trained policy, we compared it with flight errors along the set trajectory in real scenarios. The experimental results show that this system can complete the trajectory tracking tasks in the EAST vacuum vessel.