ROS-based SLAM and Path Planning for Autonomous Unmanned Surface Vehicle Navigation System

Abstract

This study addresses the development of SLAM based autonomous navigation, path planning and collision avoidance systems for the Heron unmanned surface vehicle (USV). Exteroceptive sensors including Velodyne 3D VLP32 lidar, Axis pan-tilt-zoom (PTZ) camera and omni directional camera are installed onboard to provide sensing and perception capabilities to the vessel. The development of SLAM based autonomous navigation and path planning algorithms is based on Robot Operating System (ROS) navigation stack which provides a framework for hardware and software integration including communication between processes over multiple machines. The SLAM technique utilizes the Rao-Blackwellized Particle Filter (RBPF) occupancy grid mapping algorithm to track the vessel trajectories. Next, a path is planned based on occupancy grid map obtained from SLAM in which the trajectory is designed based on finding the shortest and safe route during maneuvering. Experimental programme is conducted to verify the feasibility of the developed autonomous navigation algorithms under several scenarios, and the possibilities and challenges for safe USV autonomous navigation are also discussed. The results suggest that the USV can navigate smoothly with surrounding wind velocity of 2m/s and wave height of 0. 2m to 0. 4m.