QRKM

We connect the physical world to intelligence.

Scroll to travel through a port — from the quay to the edge, the fog and the cloud.

FILM00 / 9
Scroll

We design and build the intelligent systems that run ports

QRKM: cyber-physical architecture from endpoint to edge to fog to cloud — with AI in every layer, real-time 3D monitoring of the whole port, and foundation models fine-tuned for your operation.

QRKM Port Twin · real-time 3D
QRKM Port Twin — real-time 3D operations view

Systems for the operations where physical work and data meet

Ports & terminalsShipping linesBulk & rail operatorsLogistics & forwardersIndustrial plantsOrganisations with paper archivesPorts & terminalsShipping linesBulk & rail operatorsLogistics & forwardersIndustrial plantsOrganisations with paper archives
1.0Architecture

One system, four layers, AI in each

We are system architects and implementers. Trackers, endpoint devices, HMI monitors, edge gateways, the twin and the BI dashboards are parts of one design — not products sold one by one.

Layer 02

Edge

Compute where the work happens. Operators use HMI monitors that run models locally — querying thousands of containers and bulk trailers without waiting on the cloud.

data up models down
Connected through IoT and industrial protocols
MQTTOPC UAModbusLoRaWANNB-IoT / LTE-MLTE / 4G / 2GBluetooth LEGPS / GNSSREST & WebSocket APIs
2.0Intelligence

Cyber-physical systems, connected to foundation models

Transformers, deep learning, statistical and fuzzy systems — chosen for each decision, fine-tuned for each use, and deployed wherever the decision has to be made.

Fine-tuned model · local · connected to twin, sensors, documents
Which reefer containers on berth 2 are above their set point, and what is the risk to the vessel departure?
3 containers are above set point (+2.1 °C to +3.4 °C), all in block C4.Delay risk: low–medium (fuzzy score 0.38) — STS-02 can reach them before cut-off.Sensors · 12 s agoTwin · block C4Stowage plan
Ask the port…

Transformers

Foundation models & LLMs

Large language and vision models connected to live operational data. Retrieval over your own archive, local deployment, and agents that act inside the system — with a person in the loop.

Adaptation

Fine-tuning for specific uses

Foundation models adapted to a port’s vocabulary, documents and rules — from small task models on the edge to adapters on open-weight LLMs.

AI

Reasoning

Fuzzy systems

Decisions where rules are vague and data is noisy: congestion, priority, risk — expressed the way operators actually think.

lowmediumhigh
TrOCR and layout detection on a bill of lading

TrOCR · VLM

Transformers that read and see

TrOCR for handwriting and print, vision-language models for photos, drawings and screens — every block, line and word located.

Time series

Deep learning on sensor data

Forecasting arrivals, queues and throughput; detecting anomalies in vibration, temperature and movement.

forecast

Inference

Statistical models

Estimation, uncertainty and confidence for every number the system reports — so decisions rest on more than a single prediction.

95 % interval

Operations research

Optimisation

Berth, crane, yard and gate planning that turns live data into better moves, not just better charts.

TinyML · edge inference

Endpoint & edge AI

Models compressed to run on trackers and HMI monitors, deciding locally in milliseconds.

AI
Layer
Intelligence running there
01Endpoint
On-device models (TinyML)Handling-gesture recognitionAnomaly detection
02Edge
Real-time inferenceLocal LLMs in the HMIComputer vision
03Fog
Sensor fusionStatistical & fuzzy decision modelsYard, berth and gate optimisation
04Cloud
Foundation-model fine-tuningDocument AI (TrOCR, LLM × OCR)Fleet-wide forecasting
The intelligence layer
3.0Port operations

Real-time 3D monitoring, and the optimisation behind it

Every vessel, crane, container, bulk trailer and truck in one live 3D model — and models that turn what it sees into better berth, yard and gate decisions.

Berth plan · next 24 h

Re-planned by the model from live AIS, crane health and yard occupancy

PlannedMoved by optimiserBulk
00:0004:0008:0012:0016:0020:0024:00
B1
MSC Aurora
CMA Tiber
Ever Lumen
B2
Ocean Grain
Maersk Elba
ONE Harbor
B3
Hapag Rhine
Iron Coast
B4
Feeder 12
Yang Ming 7
COSCO Pearl

Berth utilisation

87%+6.4

Crane moves / h

31.2+2.1

Truck turn time

38 min−9

ETA forecast error

±22 min−41%

01

Berths & vessels

Line-up, ETA and progress per vessel; berth plans adjusted as arrivals change.

02

Quay cranes

Productivity per crane and predictive maintenance from vibration and temperature.

03

Yard

Stacking, utilisation and moves, optimised with live positions.

04

Gate & traffic

Truck queues and turn times forecast; lanes balanced before they jam.

05

Bulk terminal

Stockpiles, conveyors and bulk trailers measured and tracked in real time.

06

Rail

Wagons and loading in the same picture as quay and gate.

Port operations in detail
4.0Devices

Hardware designed as part of the system

We design the devices the architecture needs — from endpoint trackers to HMI monitors at the edge — so every layer speaks the same language.

HMI monitor — Port twin

Edge

HMI monitor — Port twin

Touch monitor for the operation: live map, equipment health, dashboards and 3D view of the port.

BI dashboards

Cloud

BI dashboards

Throughput, capacity, trends and forecasts for the strategic level — fed by every layer below.

Container trackers

Endpoint

Container trackers

Long-life, IP67 trackers on the container body; the smart version runs an AI model on the device.

Bulk load tracking

Endpoint

Bulk load tracking

Load, fill and route for bulk trailers, wagons and stockpiles.

5.0Delivery

From architecture to operation

One team, from the first assessment to the system running — and improving — in production. Since 2019.

  1. 01

    Architecture

    Assess the operation, define the layers, data flows and where AI belongs.

  2. 02

    Design

    Hardware, firmware, models and platforms designed together.

  3. 03

    Build

    Devices, edge software, twin and models built and tested on site.

  4. 04

    Integrate

    Connected to existing terminal, ERP and document systems through standard APIs.

  5. 05

    Operate

    Monitoring, retraining and support after go-live.

6.0Platforms & BI

Open them right here

The software layer of the system. Each platform opens in a window on this page; previews run without a server.