A Hybrid RRT*-CBF Approach for Safe Navigation of Autonomous Underwater Vehicles

Safe and efficient navigation of Autonomous Underwater Vehicles (AUVs) in complex and unknown underwater terrains requires robust online path-planning algorithms. For the successful completion of underwater missions, the design of effective collision avoidance techniques is critical to ensure the safe operation of AUVs. In this paper, an integrated approach of improved Rapidly-exploring Random Tree Star (RRT*)-based motion planning with Control Barrier Function (CBF)-based obstacle avoidance is presented for AUVs. The novelty of the work lies in employing a rolling window approach and a bias sampling method, which reduce undesirable exploration of the conventional RRT* algorithm, leading to more efficient motion planning of AUVs. Numerical simulations validate the proposed algorithm's efficacy in avoiding static and dynamic obstacles in unknown environments. The credibility of the proposed algorithm for real-time implementation of a safe AUV mission is further supported by the comparison with conventional RRT*-CBF, demonstrating its significantly reduced computational cost and improved success rate.