Vietri

NO EN

Computer vision, machine learning and industrial IoT

Vietri is the sole proprietorship of Lars Martin Ølstad. I hold an MSc from NTNU, with robotics, AI and automation as my specialisation.

Most of my work needs both code and equipment. It usually comes down to measuring something: what is happening in a process, how a machine behaves, or what a camera actually sees. I set up the sensors, collect the data, and build whatever it takes for someone to put it to use.

olstadlarsmartin@gmail.com +47 476 28 200

Services

Computer vision

Getting numbers out of video: where something is, how it moves, what angle it sits at. I work with 3D pose estimation, object detection and camera calibration, and have published research in the field.

Machine learning and AI

Models that spot anomalies, forecast what is coming, or point to the best setting. I take it all the way from raw data to a model that runs in production.

Industrial IoT

Getting the machines to talk to the rest of the system. I connect PLCs and equipment to a server over MQTT, keep the history, and build dashboards that show what is happening right now.

Electronics and mechatronics

Sensors, microcontrollers and motor control. I build measurement setups that hold up outdoors, not just on a bench in a lab.

CAD and 3D printing

I design and print parts myself. That is often the missing piece: a bracket to mount a sensor somewhere nobody has mounted anything before. If the part is made elsewhere, I design it for laser cutting or machining.

Prototyping and test rigs

Turning the sketch into something that works. I have helped build underwater vehicles, drones and measurement rigs that have been used at sea and in the air, and I always measure against ground truth before calling it done.

Selected work

Master's thesis

TrollFish — video analysis and debriefing

NTNU TrollLABS, for the Norwegian Sailing Federation and Olympic Training Centre · 2025–2026

A tool for going back through a sailing session afterwards. Video, GPS and sensor data sit on the same timeline, so a coach and athlete can jump to a particular tack and see what happened.

The system finds the manoeuvres on its own. It reads sail shape and body position from the one camera the rules allow on board, works out the wind from the GPS track, and collects everything into a report the coach can pass on.

Two peer-reviewed papers came out of the work.

TrollFish is not tied to sailing. It can be adapted to other boat classes, other sports, or somewhere else entirely where video and sensor data belong together. Get in touch if that sounds relevant.

Read more about the thesis →

An ILCA dinghy sailing at speed on open water, with mountains behind.
An ILCA in training. This is the class the system was built for.
Screenshot from TrollFish: a GPS track on a map, three synchronised video panes and a timeline showing speed, heading and heel.
Video, GPS and sensor data on one timeline.

Consulting

Reime Biotek — biogas plant monitoring

Industrial IoT and energy optimisation · 2026

Real-time monitoring of a biogas plant. Readings are pulled from the PLC over MQTT and stored as they arrive, and the dashboard shows where the plant stands right now: reactor temperature and level, gas flow to the burner, and whether the pumps and valves are doing what they should.

The process view draws the whole plant with each reading where it belongs. That lets you read a value next to the thing it describes, and see which way the medium flows and which valves are open.

I also worked on energy optimisation: when it pays to produce hot water and electricity, based on electricity prices, past consumption and what the process is doing.

Dashboard with key figures for reactor temperature, reactor level, gas flow and water store, graphs for the last 60 seconds, and a status list for the PLC, burner, pumps and alarms.
The dashboard, with key figures, history and equipment status.
Process view of the biogas plant: reactor, biomass intake and outlet, valves, circulation pumps, accumulator tank, gas burner and emergency cooler, each with its measured values.
The process view, with the sensor readings placed in the plant. The screenshots are from the demo setup, with simulated values.

Background

  • Sep – Dec 2026 Reime Biotek
    Consultant in industrial IoT. Connected the plant from PLC to server, built dashboards for real-time monitoring, and worked on scheduling hot water and electricity production around the electricity price.
  • 2025 – 2026 NTNU TrollLABS
    Master's thesis for the Norwegian Sailing Federation and Olympic Training Centre. Sensors, dashboards, machine learning and computer vision, put together into a tool the coaches could use.
  • Summer 2025 Kongsberg Discovery
    R&D engineer. Worked on getting a semi-autonomous ROV to operate from an unmanned vessel: launch and recovery, handling the tether, and remote control. Also designed and produced parts for an autonomous fixed-wing drone.
  • 2023 – 2024 Vortex NTNU
    Deputy mechanical team leader. Helped lead the development of an ROV that won an international student competition.
  • Summers 2017 – 2024 Mester Grønn
    AI developer. Built a machine learning pipeline that finds embezzlement in the stores, from collecting transaction data to a model that tells real irregularities apart from ordinary variation. Also moved the APIs from SOAP to REST.
  • 2021 – 2026 NTNU
    MSc in Engineering and ICT, specialising in robotics, AI and automation. I spent one year of the degree in Naples, at Università di Napoli Federico II.

Publications

Contact

Send an email or give me a call, and we can talk through what you need.

olstadlarsmartin@gmail.com
+47 476 28 200

Vietri Lars Martin Ølstad
Skansen 16, 3409 Tranby, Norway
Org. no. 938 483 396