3D Path Planning of Autonomous Underwater Vehicles Using a Rapidly-exploring Random Trees Algorithm

Abstract

With the growing interest in ocean exploration, research into Autonomous Underwater Vehicles (AUVs) is attracting more and more attention. Compared to land robots, AUVs have to endure complex underwater environments and take into account various factors such as tidal currents, water pressure, the topography of the environment, etc. Faced with this diversity of applications, the field of AUVs involves several research themes, particularly those relating to planning trajectories in unknown or partially structured oceanic environments. This paper proposes the implementation and evaluation of an efficient sampling-based path planning technique, namely the Rapidly-exploring Random Tree (RRT). Various performance criteria are considered to evaluate the investigated RRT-based AUV's path planning algorithms in terms of path generation and collision avoidance capabilities. Demonstrative results and comparisons are carried out over increased complexity diving scenarios to show the effectiveness and practicability of these sampling-based path planners.