Unmanned Surface Vehicle Docking Using Optical Flow and Global Positioning System

One of the challenging requirements of an Unmanned Surface Vehicles (USVs) is providing a high precision control performance during docking to prevent collisions with the dock walls. While GPS-based navigation is highly effective for guiding USVs from one location to another, it becomes less effective when the USV moves slowly and requires high accuracy. In this paper, we propose a navigation method for USVs using optical flow in conjunction with GPS. GPS remains as the main navigation source as the USV approaches the dock, but during the docking stage, the GPS is disengaged in favor of optical flow as the primary guide. We have successfully implemented this method with an error of 1.3m compared to the GPS-based has an error of 8.3 m. It means that the proposed method improves accuracy up to 7m.

I. Introduction

The rapid advancement of autonomous vehicles, including autonomous cars [1] [2], autonomous robots [3], autonomous aircraft [4], autonomous trains [5], and autonomous ships [6], is currently experiencing significant growth. Autonomous vehicles facilitate self-operated locomotion without human intervention. These systems offer several advantages, such as error reduction, consistency, accuracy, and seamless integrated operation [7]. The success of autonomous vehicle development is attributed to human achievements in the development of critical components, including the Global Positioning System (GPS), Inertia Measurement Unit (IMU), cameras, LIDAR, radar, sonar, computer vision, and artificial intelligence [8]. In this paper, we present a novel docking method for autonomous ships utilizing GPS and computer vision technology.