Fujitsu Zinrai trial speeds vessel warnings by about two minutes

AI system

System: Fujitsu Human Centric AI Zinrai collision-risk prediction using AIS and radar data
Model: Not assessed
Release: Exact release not established.
Application developer: Fujitsu Laboratories / Fujitsu Limited
Role: The system estimated collision risk from vessel-position and radar data.
Configuration: The deployment used Fujitsu Human Centric AI Zinrai.
Inputs and outputs: Not assessed
Autonomy: Not assessed
Workflow: A VTS center with radar and AIS feeds and archived risk recommendations. Replayed near-miss scenarios in a training environment. Experienced controllers comparing conventional and AI-supported operation.
Human oversight: Not assessed
Relation to AI: The AI produced risk information that controllers used to decide warnings. The two-minute gain and doubled warning count were measured against conventional operation in replayed near-miss scenarios with six controllers.
AI contribution: The AI produced risk information that controllers used to decide warnings. The two-minute gain and doubled warning count were measured against conventional operation in replayed near-miss scenarios with six controllers.

Studies: Not assessed
Study design: Not assessed
Measurement: Not assessed
Setting and sample: Not assessed
Measures: the number of warnings issued to risk-prone vessels nearly doubled
Comparison: conventional control operation without AI risk information, in a simulation of past near-miss cases
Results: Reference: conventional operation (no absolute figure given). Result: risk warning sent an average of approximately two minutes earlier; number of warnings issued for risk-prone vessels nearly doubled.
Additional value from AI: Not assessed
Limitations: Not assessed
Denominators: Not assessed
Materials for replication: Not assessed

Participants: Developer: Fujitsu Laboratories / Fujitsu Limited. Operator: Japan Coast Guard, Tokyo Wan Vessel Traffic Service Center (trial).
Jurisdictions: Developer: Japan. Operator: Japan.
Ownership and control: Not assessed
Dependencies: Not assessed
Cross-border connections: Not assessed
Geopolitical implications: Not assessed

Conditions for replication: Risk predictions are computed in real time on live traffic. Controllers trust and act on the predictions.
Replication attempts: 1 follow-up entries. Outcome unresolved: 1.
Later findings: 28 September 2021. Fujitsu ran a second field trial at the same center from 17 November 2020 to 2 September 2021 with an added course-following algorithm, reported about 90% fewer unnecessary alerts than the earlier technology, and planned a commercial navigation-support service from March 2022.
Availability: Not assessed
Institutional decision: Not assessed
Grounds: Not assessed
Action and result: Not assessed
Success criterion: Not assessed

Assessed benefits: Organizations. Society. People.
Depth: Not assessed
Reasons: Organizations: Simulation results only. No live operational outcome is documented. Society: No collision avoided or reduced is documented. People: No individual outcome is documented.
Observed or intended: Not assessed
Rubric: Not assessed
Assessor and date: Not assessed
Alternative assessments: Not assessed
Fill: Independently reported

Kind: Not assessed
Related entries: Not assessed
Field: Maritime safety. Vessel traffic services / collision-risk prediction.
Purpose: The technology was built to detect near misses and predict collision-risk areas to help vessel traffic controllers warn ships earlier.
Dates: Date basis unassessed: 2020 (precision unassessed). Updated: 21 October 2026.
Stage: Not assessed
Version: SHA-256: 033c4a9735cf…