Path Planning Scheme for AUV Based on Improved Model Predictive Control

In this paper, a path planning scheme for Model Predictive Control (MPC) of autonomous underwater vehicle (AUV) is proposed based on conditional penalty function. Traditionally, MPC-based path planning schemes use the squared difference between the current and target positions as the cost function, and consider the size of the obstacle plus a minimum safety radius as constraints. However, such methods may lead to the AUV closely following the minimum safety radius, leaving insufficient time to react to unexpected situations. To address this issue, a soft constraint is introduced into the cost function using a conditional penalty function. This allows for a relative safer distance between the AUV and obstacles to be maintained in certain situations. Simulation results demonstrate the effectiveness of the proposed scheme.