Riverbank Following Planner (RBFP) for USVs Based on Point Cloud Data

Abstract

Autonomous path planning along riverbanks is crucial for unmanned surface vehicles (USVs) to execute specific tasks such as levee safety detection and underwater pipe inspections, which are vital for riverbank safety and water environment protection. Given the intricate shapes of riverbanks, the dynamic nature of tidal influences, and constraints in real-time cartographic updates, there is a heightened susceptibility to inaccuracies during manual waypoint designation. These factors collectively impact the efficiency of USVs in following riverbank paths. We introduce a riverbank following planner (RBFP) for USVs to tackle this challenge. This planner, utilizing 2D LiDAR, autonomously selects the following point to follow riverbank shapes. Additionally, a PID controller is integrated to compensate for position and yaw errors. Our proposed method reduces the deviation between the USV’s planned path and the actual riverbank shape. We simulated straight, convex, and concave riverbanks in the Virtual RobotX (VRX) simulator while considering the impacts of wind, waves, and USV dynamics. The experimental result indicates the following performance of 96.92%, 67.30%, and 61.15% for straight, convex, and concave banks, respectively. The proposed RBFP can support a novel autonomous navigation scenario for autonomous paths following along the riverbank without any preplanned paths or destinations.