Semester Award Granted

Spring 2025

Submission Date

May 2025

Document Type

Thesis

Degree Name

Master of Science (MS)

Thesis/Dissertation Advisor [Chair]

Manhar Dhanak

Abstract

USVs play a critical role in a wide range of marine applications, where precise station-keeping is essential for tasks such as autonomous launch and recovery of an onboard aerial drone. Maintaining a stable position in a dynamic marine environment requires robust control strategies to counteract external forces, including wind, waves, and currents, which can introduce vehicle oscillations and positional errors. To optimize the operability limits of the vehicle based on sea state, weather conditions, and mission requirements, the USV must be carefully tuned. Traditionally, the tuning is performed manually, which is time-consuming and can lead to operational delays. This study explores AI-driven auto-tuning of PID controllers for station-keeping in prevailing environmental conditions, using evolutionary optimization algorithms, specifically NSGA-II and CMAES, through field trials with a WAM-V 16 USV. To assess heading and position control accuracy, the trials were performed under varying ecological conditions, including crosswind scenarios and tidal current disturbances. Results demonstrate that the AI-tuned controller is more efficient and significantly enhances the station-keeping performance compared to traditional manual tuning methods. These findings contribute to advancing control methodologies for autonomous USVs, improving reliability and adaptability in challenging marine environments.

Download

Share

COinS