Predictive Maintenance and Remaining Useful Life Prediction for Collaborative Robots in Shipbuilding
Developing predictive-maintenance and remaining-useful-life (RUL) prediction models for collaborative robots operating in shipyard manufacturing, as part of a national multimodal foundation model R&D program.
2026.03 –
·Motivation
- In shipbuilding, collaborative robots operate under highly variable, ship-specific conditions, so robot maintenance still relies on experience-based, manual judgment rather than systematic prediction.
- Existing rule-based and single-task AI/ML approaches struggle to generalize across changing designs, new ship types, and heterogeneous multimodal data (drawings, schedules, sensor logs).
·Goal
- As part of a national multimodal foundation-model R&D program (with HD Hyundai Heavy Industries, HD Korea Shipbuilding & Offshore Engineering, and Crowdworks), lead the predictive-maintenance track: anomaly detection, fault diagnosis, remaining useful life (RUL) prediction, and maintenance scheduling for collaborative robots.
- Build a digital twin of robot health that supports explainable, dialogue-based maintenance decisions.
·Methodology
- Define the state-space and data pipeline for collaborative-robot condition monitoring, then develop an adaptive physics-informed RUL prediction model that stays reliable under small data and shifting work conditions.
- Extend to a Bayesian neural network for RUL distributions and confidence intervals, feeding a multi-stage (normal/caution/warning/risk) alert system.
- In later stages, integrate diagnosis into an LLM-based conversational maintenance agent and extend to multi-robot coordination.
