Enhanced obstacle avoidance for autonomous underwater vehicles via path integral control based on guiding vector field

Abstract

Autonomous Underwater Vehicles (AUVs) face significant challenges in tracking, navigation, and obstacle avoidance—critical aspects for advancing intelligent underwater robotics. This research presents a new navigation technique that combines Guiding Vector Field (GVF) concepts with Model Predictive Path Integral (MPPI) control to improve the precision and efficiency of vectored thruster AUV operating in complex environments. The proposed approach utilizes the AUV's relative positioning and environmental data to generate obstacle avoidance trajectories as desired GVF. Subsequently, MPPI optimization is applied to control inputs, considering dynamic constraints, to achieve effective tracking and obstacle avoidance. Extensive simulation experiments demonstrate the method's efficacy in navigating complex scenarios with non-convex obstacles. In the aspect of path tracking, the tracking error is reduced by 64 %, while maintaining safe distances in various obstacle configurations. Results show that the integrated method successfully combines local optimization prediction capabilities of MPPI with the global velocity planning of GVF, enabling efficient AUV navigation in intricate environments while ensuring the effectiveness and safety of the execution process. The method demonstrates robust performance even under disturbance conditions, maintaining a tracking error of only 0.017 m. This research contributes a solution for AUV operation in challenging maritime settings, with applications in marine surveying, underwater search and rescue, and offshore operations. By addressing key challenges in underwater navigation, this study advances the practical capabilities of AUV technology, paving the way for more efficient and reliable underwater robotic systems.