Key Learning Outcome: This presentation introduces Pathfinder, a novel ROV control framework that enables seamless integration of advanced AI models with traditional control systems while maintaining robust real-time performance in subsea operations.

Traditional ROV control systems often struggle to incorporate modern AI capabilities without sacrificing performance or reliability. This presentation details the architecture of Pathfinder, a control framework specifically designed to bridge this gap in subsea operations.

The framework provides a unified platform that maintains precise low-level control while enabling high-level AI decision making through a modular architecture. Pathfinder’s design allows for real-time execution of various control strategies, from traditional algorithms to advanced AI models, while ensuring consistent performance and reliability required for subsea operations.

The framework includes integrated support for reinforcement learning implementations, enabling future autonomous capabilities through direct environmental interaction. Currently being developed alongside Nautilus, a dedicated simulation environment, Pathfinder creates a complete ecosystem for testing and deploying AI-enhanced control strategies in subsea applications.

The presentation will cover the framework’s technical architecture, integration methodology, and early implementation results from field testing. Special attention will be given to how Pathfinder’s design choices address common challenges in combining AI capabilities with traditional ROV control systems.

Co-authors: Batuhan Ergun, Ozgur Turk