PRISMA.
Funding Source
AUTOASSESS – Horizon Europe Cascade Funding,
GA ID 101120732 (Open Call)
Project Title
Predictive Risk Intelligence for Structural Maritime
Assessment
Programme
HORIZON-CL4-2022-DIGITAL-EMERGING-02-07
Duration
2026 (9 months)
URL
AUTOASSESS
Role
- Sole Beneficiary · Technology
- Development · Exploitation · Dissemination
Description
Ballast tanks and cargo holds are among the hardest spaces on a ship to inspect — vast, GNSS-denied, and brutal on drone battery life. PRISMA adds a predictive intelligence layer to AUTOASSESS's autonomous inspection drones: a Graph Neural Network trained on 10–15 years of vessel-specific thickness measurements and survey records generates dynamic 3D risk heatmaps that tell a drone where structural degradation is most likely before it even launches. By replacing indiscriminate scanning with targeted, history-informed navigation, PRISMA supports AUTOASSESS's goal of cutting dry-dock inspections from 15 days to 3 — addressing an estimated €11 billion annual cost across the global fleet.
Our Role
bSpoke designs, builds, and validates PRISMA end to end — from historical data structuring through to live operational deployment.
-
Predictive AI model development
Heterogeneous Graph Transformer with uncertainty quantification and SHAP-based explainability -
Digital Twin integration
connecting risk heatmaps into AUTOASSESS's Cognite Data Fusion environment and mission-planning layer -
Operational validation
TRL 5→7 demonstration on an operational vessel.
