ENGINE-GUARD

The ENGINE-GUARD framework is a comprehensive robotic architecture designed to execute autonomous engine room patrolling and early event detection. It bridges the diagnostic gap left by fixed static sensors by mobilizing a specialized diagnostic payload within a dynamic rail-guided patrol framework. This rack-mounted shuttle carrier achieves a competitive advantage through four core integrated innovations. First, to overcome severe external motion blur from vessel motion dynamics and structural vibrations, it utilizes dynamic vision and neuromorphic navigation via a stereo event-camera, operating on a dual-layer rail layout that seamlessly transitions from a macro-level to a close view at critical equipment. Second, it provides mobile multi-sensory diagnostics acting as a digital stethoscope, deploying thermal, acoustic, and gas sensors directly at the component level to capture localized friction, cavitation, and VOCs. Third, to overcome maritime data sparsity, the software employs an adaptable probabilistic framework that utilizes either Probabilistic Combination Networks (based on expert-defined conditional scenarios) or Probabilistic Deep Learning, scaling dynamically based on operational data availability over the project timeframe. Finally, the system features true operational autonomy driven by Event-Triggered Rerouting and Autonomous Adaptive Sensing; upon detecting an initial anomaly or an alert from local static equipment sensors, the platform autonomously overrides its mission, calculates the shortest path to the hazard, physically aims its payload, and cross-correlates data streams through a cascading verification matrix to definitively confirm degradation and eliminate false alarms.