Predictive Dynamic Collision Avoidance of Unmanned Surface Vehicles Using Behavior-Based Method in Maritime Environment

EasyChair Preprint 4785

Abstract

A good ability of the dynamic collision avoidance (DCA) is essential for the unmanned surface vehicle (USV), which is the focus of this paper. Since current research mainly uses real-time navigation information to achieve collision avoidance to other vessels, however, in the realistic maritime environment, USV can rarely obtain such real-time information through automatic identification system (AIS) or other equipment. So, in this paper, a Kalman filter-based predictive dynamic collision avoidance of unmanned surface vehicles is proposed using the behavior-based method. The Kalman filter (KF) is integrated into the USV planner to predict the trajectories of other obstacle ships and several behaviors are designed in the light of the International Regulations for Preventing Collisions at Sea (COLREGs) to implement collision avoidance. Simulations involving three moving obstacle vessels with changing navigational statuses are presented and realistic broadcasting intervals of a class A AIS device are used in the simulation to indicate that the Kalman filter can reasonable predict the positions of moving obstacle ships and the USV can effectively make obstacle avoidance behaviors that meet the requirements of the COLREGs.

Keyphrases: COLREGS, Kalman filter, dynamic collision avoidance, unmanned surface vehicle

@booklet{EasyChair:4785,
  author    = {Jin Zou and Guoge Tan and Lei Wan and Jiayuan Zhuang},
  title     = {Predictive Dynamic Collision Avoidance of Unmanned Surface Vehicles Using Behavior-Based Method in Maritime Environment},
  howpublished = {EasyChair Preprint 4785},
  year      = {EasyChair, 2020}}