🔔 Subscribe to ShipUniverse Weekly →
Maritime artificial intelligence is moving into a different phase in late September 2026. Recent projects are increasingly tied to operating fleets, ports and shipyards rather than standalone demonstrations: CIDO Shipping has committed 20 new PCTCs to Avikus autonomous-navigation technology with another 12 vessels contemplated; Eastern Pacific Shipping is expanding Orca AI after a five-ship pilot covering more than 130,000 nautical miles; YILPORT says optimization software lifted yard-crane productivity 21% at Gebze; India's JNPA has awarded a ₹92.95 crore AI/ML digital-twin project covering roughly 34 square kilometers; HD KSOE has received approval in principle for AI to autonomously optimize auxiliary machinery; and Hanwha Ocean has contracted development of autonomous welding technology. At the same time, the IMO's first MASS Code has been in effect since July 1, establishing a regulatory route for autonomous commercial ships while retaining human accountability.
The industry is moving from “AI can advise” toward “AI can act”
The important change is not a sudden arrival of crewless ships. It is the gradual transfer of bounded operating decisions to software: route and RPM control, machinery optimisation, crane dispatch, digital-twin decision support and adaptive shipyard robotics.
Where maritime AI is actually entering operations
The deployments are no longer concentrated on one bridge screen. AI is spreading vertically through the ship and horizontally across the transport chain.
Collision avoidance, route following, RPM optimisation, watchkeeping analysis and bridge-event detection.
HD KSOE's VesselWise is moving from concept approval toward validation aboard a 174,000-cbm LNG carrier.
Crane and yard software continuously evaluates demand and recommends equipment allocation while human dispatchers retain authority.
AI/ML models are beginning to connect vessel traffic, gates, terminals, road flows and infrastructure into one decision layer.
Vision-driven welding development aims to let robots recognize joints and adapt rather than simply repeat fixed programmed paths.
Procurement, invoice matching, compliance, voyage analytics and safety-event review are increasingly AI-assisted.
Dashboards, predictions, alerts and recommendations. Humans interpreted the result and performed the action.
AI is beginning to directly adjust heading, RPM, machinery settings and equipment deployment inside tightly defined operating envelopes.
Maritime AI: From Demo to Deployment
| Project | AI / Automation Layer | Current Scale | Evidence So Far | Human Role | Current Status |
|---|---|---|---|---|---|
| Avikus + CIDO Shipping | Navigation, collision avoidance, route following and RPM control. | 20 PCTCs 7,600 CEU each, with another 12 vessels contemplated. | Fleet contract signed. Ships are under construction and deliveries begin from 2027. | Bridge crew remains responsible and can take manual control. | Contracted Deployment |
| Orca AI + Eastern Pacific | AI watchkeeping, perception and navigational risk analytics. | 130,000+ nm Five-vessel pilot followed by rollout to dozens more vessels. | Company reports close encounters per 1,000 nm down 48% and average minimum passing distance up 10%. | System assists bridge teams rather than replacing the watch. | Scaling After Pilot |
| YILPORT + Kaleris | RTG allocation and yard-equipment optimization. | 31 RTGs Operating at Gebze Terminal. | YILPORT and Kaleris report a 21% increase in moves per hour within three months. | Software recommends assignments; dispatchers retain control. | Operational |
| JNPA + SoftTech | AI/ML digital twin and integrated port decision support. | ₹92.95 Crore Roughly 34 km² of port operations. | Awarded project will connect seaside, landside, road traffic and a 24/7 command-and-control center. | AI supports planners and operational controllers. | Awarded |
| HD KSOE VesselWise | Autonomous auxiliary-machinery optimization. | 174,000 cbm LNG carrier planned for future onboard verification. | Lloyd's Register has issued Approval in Principle against safety, performance and security criteria. | Control architecture retains defined safety boundaries. | AiP / Validation Next |
| CMA CGM + SDARI + BV | AI-enhanced assisted containership design and decision support. | 1 JDP Technology-neutral containership concept. | Project will study efficiency, safety, workload reduction and the split between automation and crew control. | Human oversight is an explicit project design requirement. | Concept Development |
| Fugro Singapore | AI marine Geo-data interpretation and digital twins. | APAC Hub Regional development from Singapore. | Platform is intended to unify geophysical, geotechnical, laboratory and imaging datasets. | AI accelerates interpretation and predictive analysis. | Under Development |
| Hanwha Ocean + MaumAI | Vision-guided autonomous shipyard welding. | KRW2.106B Contract runs through July 31, 2027. | Development aims at adaptive welding intelligence rather than conventional fixed-path automation. | Engineers and production teams supervise the robotic process. | Development Contract |
| GTT Marine / Danelec | Automated bridge-event analytics using voyage recorder data. | Commercial BOQA launched September 1. | Software automatically flags safety-relevant operating events from routine voyages for further review. | Fleet safety teams investigate and interpret flagged events. | Available Now |
| IMO MASS Code | Regulatory framework for remotely operated and autonomous ships. | Global Non-mandatory Code effective July 1, 2026. | Covers approval, software, navigation, connectivity, remote operations, cybersecurity and human responsibility. | Master retains overall responsibility under the Code. | Regulatory Framework Live |
Respondents committing more than 10% of R&D budgets to autonomy, up from 34% in 2025.
Say market penetration is below expectations, up from 44% a year earlier.
Identify Asia as the autonomy frontrunner in the 2026 survey.
Expect autonomous vessels to exceed 10% of the commercial fleet by 2040.
Maritime AI Payback & Scale Screen
Test whether a proposed AI deployment can clear a basic commercial hurdle using your own operating assumptions rather than vendor projections.
Under these assumptions the deployment creates positive annual operating value, but the initial integration cost is not recovered inside three years.
We welcome your feedback, suggestions, corrections, and ideas for enhancements.
Please click here to get in touch