Shipboard Technology for Engine Work, Autonomous Repair, and Diagnostics (STEWARD)

The U.S. Navy's plan to expand its surface fleet with Unmanned Surface Vessels (USVs) will need abilities to function autonomously for extended periods.  Specifically, the heart of the operation, the diesel engine(s), will need special attention for providing maintenance tasks, identifying issues, and even making repairs.  The technology for providing such extended duration operations without humans-in-the-loop requires a fusion of proven reliable hardware and cutting-edge learning and training models for the system ability to autonomously identify, act, and even repair systems on and around the onboard engine. Journey Robotics, in partnership with Carnegie Mellon University’s Robotics Institute, Blue Water Autonomy, and RPM Industries, presents the Shipboard Technology for Engine Work, Autonomous Repair, and Diagnostics (STEWARD) system. STEWARD is an advanced robotic solution designed to enhance USV operational reliability through predictive maintenance and autonomous repairs. The system integrates machine learning (ML), computer vision, and a digital twin framework to monitor engine performance, detect anomalies, and execute maintenance tasks that are critical for long-duration missions without human presence. A key innovation is STEWARD’s physics-enabled digital twin, which enables real-time monitoring and predictive maintenance. Using reinforcement learning-based manipulation planning, the system autonomously inspects, diagnoses, and resolves engine-related issues. The digital twin also allows for offboard system training, enabling operators to refine ML models remotely for new autonomous repair behaviors.STEWARD features an onboard robotic arm with cameras, torque sensors, and specialized tools for inspections and repairs in confined shipboard environments. The robot operates within a gantry system for precise movement, ensuring accessibility to critical engine components. A robust sensor suite, including thermal and borescope cameras, helps detect leaks, cracks, and wear. The system follows a "Human ON the Loop" approach, where operators validate corrective actions rather than directly controlling robotic movements. If a scenario exceeds the system's capabilities, the operator can guide a solution via the digital twin and train the model using human demonstrations, enhancing adaptability over time. Phase I focuses on refining STEWARD’s design, integrating digital twin technology, and developing an initial proof-of-concept system for a Medium Unmanned Surface Vehicle (MUSV) equipped with a Caterpillar C18 Genset module. Testing in a simulated maritime environment will validate its resilience to shock and vibration. By enabling autonomous engine maintenance, STEWARD enhances the Navy’s ability to sustain long-duration, unmanned operations, ensuring mission success with minimal human intervention. Journey Robotics and its partners are committed to pioneering this transformative technology, shaping the future of autonomous maritime operations.