Robust decision making for reactive collision avoidance of maritime autonomous surface ships

Lee, Paul

Thesis

Robust decision making for reactive collision avoidance of maritime autonomous surface ships

Download PDF

Creator

Lee, Paul

Rights statement

Strathclyde Thesis Copyright

Awarding institution

University of Strathclyde

Date of award

2024

Thesis identifier

T17068

Person Identifier (Local)

202074883

Qualification Level

Doctoral (Postgraduate)

Qualification Name

Doctor of Philosophy (PhD)

Department, School or Faculty

Department of Naval Architecture, Ocean and Marine Engineering

Abstract

The maritime industry is expected to utilize maritime autonomous ships (MASSs)to realise new sustainability levels, but their ability to make robust decisions withouthuman intervention during safety-critical operations, such as collision avoidance,remains a challenge.This research aims to propose a holistic methodology that enhances the robustnessof decision making for the reactive collision avoidance of a MASS from design tooperation. In the first phase, the risk-informed design of the collision avoidancesystem (CAS) is conducted to reduce critical internal collision risk indices (CRIs). Inthe second phase, a digital twin environment is developed to simulate critical collisionscenarios pertaining to internal CRIs-based dynamic safety domain, propeller-ruddercoupled control, noisy perception sensor, critically proximate and unclassified staticor dynamic obstacles. In the third phase, a decision-making agent is optimisedfor the reactive collision avoidance operation within the above scenarios in a deepreinforcement learning (DRL)-based training framework. In the fourth phase, aclassification agent is optimised for the identification of static or dynamic obstaclesbased on real-time perception sensor data in a deep learning training framework. Inthe fifth phase, the robustness of decision-making is quantified, verified, and validatedwithin and outside the investigated scenarios.Fault tree analysis, deep deterministic policy gradient algorithm, and feedforwardneural networks are employed for the risk analysis, DRL-, and DL-based trainingframeworks, respectively. A high-speed single screw container ship and a light detection and ranging (LIDAR) sensor are considered for the reference ship. The resultsdemonstrate that a risk-informed design conducted from a cost-benefit perspectiveleads up to 91% reduction of the CAS probability of failure from its baseline configuration. Robustness of decision-making agent for the reactive collision avoidanceis verified even when the safety domain and LIDAR noise is increased by 20% and132% from its maximum trained value, respectively, and the size of the criticallyproximate static obstacle and size of the static or dynamic obstacle is increased by200% and decreased by 88.5% from its trained value, respectively. The ability of thedecision-making agent to prioritise safety over efficiency depending on the size of the safety domain and LIDAR noise level, conduct sophisticated propeller-ruddercoupled control manoeuvres when encountered with critically proximate static obstacles, and conduct efficient or navigational rules-compliant evasive manoeuvreswhen encountered with static or dynamic obstacles are noted. The accuracy of theclassification agent is verified above 98.12% even in different LIDAR configurations.The novelty of this research stems from the introduction of a framework thatenhancing the safety against the unknown-unknowns by increasing the robustnessagainst critical CRIs, such as system probability of failure, uncertainty of sensormeasurements, collision-inducing obstacles, and limited information. This researchcontributes towards the development of safe maritime autonomous systems that arecapable of making robust decisions under safety-critical scenarios.

Advisor / supervisor