Soft Actor-Critic Based Controller for Offshore Lifting Arm System of Active 3-DOF Wave Compensation Using Gym Environment with PyBullet

Offshore lifting arm system (OLAS) is widely used in many types of offshore lifting operations. The OLAS motions in heave, roll and pitch, induced by waves, threaten the safety and accuracy of the OLAS operation. This paper proposes a deep reinforcement learning (DRL) approach to address the challenge. First, a Gym environment for modeling OLAS dynamics with wave-induced motion using PyBullet is established. Then, a novel soft actor-critic (SAC) based active 3-DOF wave compensation controller is proposed. It utilizes a multi-stage incentive reward function to enhance exploration performance. Simulation results demonstrate the SAC based controller's significant advantages over the proportional-differential (PD) controller in terms of compensation efficiency and vibration suppression.