Linker Vision ← All products / Port

Port

Port Digital Twin & Simulation

A photoreal port digital twin is built with 3D Gaussian Splatting and brought into NVIDIA Omniverse, fusing five data sources — CCTV, AIS, port scheduling, met-ocean data and the vessel database — into a single platform for predictive simulation and scenario display. The system runs on a five-layer architecture (sensing, data, model, simulation, presentation) and, once virtual and physical are aligned, holds measurement error within 2%, survey-verified point by point at 20 or more locations. It supports three core simulations — berthing and unberthing trajectory prediction, berth-fit assessment, and whole-field view reconstruction for navigation — and lets camera layouts, emergency procedures and what-if decisions be rehearsed in the virtual environment first. Port management moves from after-the-fact discovery and manual reporting to predictive simulation and real-time notification.

Customer challenges

1

Port information sits in separate AIS, CCTV and vessel-dynamics systems that are never aligned, forming data silos

2

Existing CCTV is not registered against vessel dimensions or channel coordinates, so spatial distance cannot be quantified

3

With tall bridge structures and shore poles only about 10 metres high, masters and pilots face significant blind spots

4

There is no systematic way to judge automatically whether a berth can take an arriving vessel

5

Rare hazardous events (collision, man overboard, spillage) leave almost no real data to train models on

6

Camera layouts, emergency procedures and investment decisions can only be trialled on the live site, at cost and risk

7

No objective, measurable, replayable spatial baseline exists, so incidents cannot be reviewed nor pre-empted

Customer benefits

Port information scattered across systems is aligned in one 3D scene, eliminating data silos

Imagery carries real-world coordinates, so distance is measured rather than estimated by eye

Navigation blind spots are covered by whole-field reconstruction, giving pilots and masters objective grounds

Synthetic data supplies training samples for rare hazards without waiting for an incident

Camera layouts and emergency procedures are rehearsed virtually, cutting the cost of on-site trial and error

Measurement error within 2% provides the spatial baseline for setting safety thresholds and reviewing incidents

The same twin extends to pilot AR guidance, unmanned port patrol and validation of automated berthing

Key features

3DGS Port Reconstruction (Quays, Channel, Berthing Waters, Turning Basin)

Reconstructs quays, channel, berthing waters, and turning basin as a 3DGS port twin.

Alert Threshold Calibration (Kinetic Energy Ratio vs Fender Design Capacity)

Calibrates alert thresholds by comparing kinetic energy ratio against fender design capacity.

Rare-Scenario Simulation (Night, Sea Fog, Strong Wind, Swell, Backlight)

Simulates rare scenarios such as night, sea fog, strong wind, swell, and backlight.

Camera Placement Optimisation & Blind-Spot Analysis

Optimises camera placement and analyses blind spots.

Berth-Fit Assessment & Whole-Field View Reconstruction

Assesses berth fit and reconstructs a whole-field view.

Berthing Trajectory Simulation (Vessel Type, Tonnage, Speed, Angle, Wind & Current)

Berthing Trajectory Simulation (Vessel Type, Tonnage, Speed, Angle, Wind & Current)

Synthetic Data Generation (Small Non-AIS Craft, Long-Range Channel, Adverse Weather)

Synthetic Data Generation (Small Non-AIS Craft, Long-Range Channel, Adverse Weather)

Before & after

Before Information scattered across systems, cross-referenced by hand After — with Linker AI Nexa Five sources aligned on one timeline and in one 3D scene
Before Imagery judged by eye, with no way to quantify distance After — with Linker AI Nexa Measurable once registered, with error ≤ 2%
Before Blind spots covered by experience alone After — with Linker AI Nexa Multi-angle fusion rebuilds the overhead view
Before No data available to train on rare events After — with Linker AI Nexa Synthetic training data generated in the twin
Before Layouts and procedures trialled only on the live site After — with Linker AI Nexa Rehearsed virtually before anything is built
Before Only recollection and scattered footage for review After — with Linker AI Nexa Replayable tracks and spatial records