Research on Control Problems of vectored thruster AUVs with deep reinforcement learning

In comparison to traditional fin and rudder propulsion devices, vector thruster autonomous underwater vehicles (AUVs) are renowned for their superior maneuverability, particularly at low or zero speeds. This study introduces a control strategy for vector thruster AUVs that employs deep reinforcement learning algorithms for three-dimensional random point and trajectory tracking. Simulation tests are conducted to investigate the motion of vector propelled AUVs. The utilization of deep reinforcement learning algorithms in this research demonstrates the potential of this approach in designing control strategies for vector thruster AUVs, ultimately enhancing maneuverability and autonomy in underwater exploration.