Safe and Predictive Path-Planning of Autonomous Underwater Vehicles in a Dynamic Cluttered Environment

Abstract

In underwater navigation, precise waypoint tracking and reliable obstacle avoidance are essential for the safe and efficient operation of Autonomous Underwater Vehicles (AUVs). In particular, cluttered and dynamic environments pose a challenge for safe navigation. This paper presents a novel control framework that combines Nonlinear Model Predictive Control (NMPC) with Collision Cone Control Barrier Functions (C3BFs) to enable robust and real-time dynamic obstacle avoidance for slender-body AUVs. These vehicles face unique challenges due to underactuation and actuator constraints. To address this, a regularization term is incorporated into the NMPC cost function to ensure smooth control transitions and compliance with actuator limitations. The C3BFs are formulated to maintain safe distances from moving obstacles by enforcing collision-free trajectories based on predicted relative motion. The combination of NMPC-C3BF ensures safe and reliable navigation in challenging underwater environments. Extensive simulations show that the proposed framework allows the AUV to successfully avoid dynamic obstacles while still following its planned path accurately. © 2025 Elsevier B.V., All rights reserved.