An automatic COLREGs-compliant obstacle avoidance system for an unmanned surface vehicle

Abstract

Unmanned surface vehicles are becoming increasingly vital tools in a variety of maritime applications. Unfortunately, their usability is severely constrained by the lack of a reliable obstacle detection and avoidance system. In this article, one such experimental platform is proposed, which performs obstacle detection, risk assessment and path planning (avoidance) tasks autonomously in an integrated manner. The detection system is based on a vision-LIDAR (light detection and ranging) system, whereas a heuristic path planner is utilised. A unique property of the path planner is its compliance with the marine collision regulations. It is demonstrated through hardware-in-the-loop simulations that the proposed system can be useful for both uninhabited and manned vessels.

COLREGs-Compliant Path Planning for Autonomous Surface Vehicles: A Multiobjective Optimization Approach

“…Typical techniques include artificial potential fields (Naeem et al (2016)) and heuristic A * method (Campbell et al (2014)) developed by the authors' research group, velocity obstacle method (Kuwata et al (2014)) and Evolutionary algorithms (Szlapczynski (2011)). However, most, if not all of the existing techniques do not scale well to multiple target ships and multiple COLREGs rules such as rules 2, 8, 13-17 in Cockcroft and Lameijer (2003), and usually one objective can be considered only when using these techniques.…”

“…Note that this part of the system is only activated and will be continuously building situational awareness picture and assessing the risk, only if a target/obstacle is detected. To assess a risk of collision, the widely-used closest point of approach (CPA) method has been adopted for evaluating if there is a potential collision risk in the near future (Campbell et al (2014); Bertaska et al (2015); Kuwata et al (2014)). Briefly speaking, this method compares the time to closest point of approach (TCPA) and the distance to closest point of approach (DCPA) with prescribed parameters t max and d min , where t max and d min are dependent on the vessel type and also the environment where she is being operated.…”

An Automatic Navigation System for Unmanned Surface Vehicles in Realistic Sea Environments

“…Although these algorithms have good real-time performance, they are difficult to use in complex and irregular environments. To solve this problem, some scholars use the hierarchical structure to combine global path planning and local path re-planning [6,11,12], and some try to improve the traditional path planning algorithm, using for example Rule-based Repairing A* [8]. Liu et al selected the fast marching method (FMM)-based path planning algorithm [20].…”

“…To date, the collision avoidance control algorithm has been developed in mobile robots [23], unmanned vehicles [24,25], and unmanned aerial vehicles (UAVs) [26], but there are few studies on USVs. A CAS for USVs emerged in relevant literature [8], in which the fuzzy estimator method was developed for collision avoidance control. The method is able to avoid collision with a change of course.…”

“…These include Dijkstra's algorithm [6], A* [7,8], Theta* [9], the Voronoi diagram [10], particle swarm optimization (PSO) [11], ant colony optimization (ACO) [12,13], the genetic algorithm (GA) [14], and so on. Recently, considering the turning capacity of USVs, Yang and Tseng used the finite Angle A* algorithm (FAA*), which is used to determine safer and suboptimal paths for USVs on satellite thermal images [7].…”