State forest management
Systematic stand planning and long rotation thinking, from foresters who managed public land for a lifetime.
Autonomous forest thinning · Estonia
Smarter forest.
Smarter forest.
Stronger future.
In harmony with nature.
REKU builds a lightweight autonomous forestry machine for thinning. It chooses which trees to remove the way an experienced forester would, and leaves the ground intact.
REKU is not a harvester.
REKU tends the forest.
The problem
Three pressures close in on thinning.
Thinning decides the value of a forest for decades. Today it is postponed or skipped, because the tools, the economics and the people behind it are all running out at once.
Ecological
~18 %of a thinned stand goes to strip roads
Heavy machines in young stands
Harvesters and forwarders of 20 tonnes or more need wide strip roads. These take about 18% of the stand and cut its production by 5–10% over the next 15–20 years. The machines also compact the soil and damage the trees that stay, and milder winters leave soils less able to carry them.
Economic
34 / 44 €cost vs price per m³
Thin margins, postponed work
A light thinning service costs about €34 per m³ against a market price of about €44. With a margin that thin, owners put the work off, and an under-tended stand yields lower-value wood for decades.
Human
60 %of the cost is operator wages
Skilled operators are retiring
The people who know how to thin well are leaving, and few follow. Their judgement has never been written down, and their wages already make up most of the cost.
Softer soils
Lighter machines
Slower work
Lower cost base
No room for a wage in the cab
Automation
These are not three separate problems. The sector has to automate thinning, whoever does it. REKU intends to be the one who does it well.
Strip roads and production loss: Cedergren, J. et al. (2024), Skogforsk Arbetsrapport 1225-2024, citing Agestam (2015). Cost, price and wage share: Susi, R. (2024), Estonian University of Life Sciences, study of Usewood thinning service on 33.7 ha.
Technology
Intelligence that understands forests.
A 2.5-tonne machine that sees the stand in 3D, decides which trees stay and which go, and carries out the work itself.
3D scan · thinning decision, illustrative
Keep
Remove
Terrain
- Platform
- Usewood Combi light machineA proven lightweight platform that both fells and forwards. One machine does the complete thinning job.
- Weight
- About 2.5 tonnesLight enough to work between remaining trees without rutting the soil.
- Vision
- LightCode Photonics 3D machine visionReads stem positions, diameters and spacing in the stand, in real time.
- Compute
- NVIDIA DGX Spark on boardRuns tree selection and movement decisions at the machine, without a network connection.
- Decisions
- Trained on expert judgementSelection follows REKU's own thinning style, learned from foresters with decades in the field.
Cab, remote-ready
Crane
Harvester head
Light bogie wheels
Usewood light forestry machine, the base for REKU
The platform
Precision over brute force.
REKU is built on the Usewood light forestry machine, a proven Estonian design. We add the eyes, the computer and the judgement. The mechanics are already tested in the forest.
- Works between treesNarrow and light, so it needs far fewer strip roads than a full-size harvester.
- Gentle on soilLow ground pressure means far less rutting and root damage than a full-size harvester.
- Moves on a car trailerFits a standard 3,500 kg trailer. No low-loader, no heavy logistics.
- Fells and forwardsOne machine cuts the selected trees and carries the timber out.
The thinning style
Forty years of field judgement, made learnable.
Good thinning is tacit knowledge. We record how experienced foresters choose trees and turn those choices into structured training data. The resulting style blends three traditions.
Private forestry
Practical choices that keep a family forest healthy and valuable, one stand at a time.
Continuous-cover forestry
Keeping the forest standing and diverse, with selective removal instead of clear-felling.
The machine should leave a stand the way a good forester would leave it. That is the bar we train against.
REKU design principle
Data from day one
Every decision linked to its outcome.
From the first field season we collect data in a form that can train world models of the forest. Three rules shape how we record it.
Rule 1
One timeline
3D sensing, machine movement and crane actions are time-synchronised, so every frame shows what the machine saw and did at that moment.
Rule 2
Action and consequence
We log what the machine did and what happened as a result, not only which tree was felled.
Rule 3
Measure again
The same area is scanned before thinning, right after, and in later years, to tie each decision to how the stand actually grows.
Service
Thinning today, autonomy next.
We start with what pays today: an experienced operator runs the REKU machine and does careful thinning at the market rate, and every hectare also produces training data. As autonomy grows, one remote operator will supervise a cluster of three machines from one screen, stepping in where a person should decide. Taking the operator out of the cab is what makes careful thinning pay.
- Complete thinning, one machine type. Each machine cuts the selected trees and brings the timber to the roadside.
- Gentle on the ground. Low weight protects roots and soil, and the remaining trees keep their bark.
- Built for Nordic winters. Frozen ground is the best season for thinning, and our machine is designed around it.
- A record of every stand. Forest owners get the scan data behind each thinning decision.
Thinning as a service
Machine as a service
Decision model licensing
For forest owners
Illustration · planned owner view
Your forest, in one view.
Every REKU job will come with a live view of your stand. You see where the machines are, what has been cut, how much timber is at the roadside and how gently the work was done.
REKU Stand overview · Pärnu County, example stand
Live · updated 2 s ago
Timber by assortment
Pulpwood60 m³
Small logs30 m³
Firewood55 m³
Timber income to date
€10,164
145 m³ at the roadside, at current buyer prices
Machines
REKU-1
Working
REKU-2
Working
REKU-3
Fuel low
Stand map
Stand boundary
Thinning corridor
Machine
Gentle work
Ground pressure22 kPa OK
Rut depth30 mm OK
Thinning intensity28 %
Area done4.1 / 6.0 ha
Every stem counted
Volumes by assortment as they are cut, ready for the timber buyer.
Proof of gentle work
Ground pressure and rut depth are measured and kept with the job record.
See it as it happens
Machine positions and progress across the stand, live, from any device.
Roadmap
Where we are, and what comes next.
We are early, and we say so. The concept is defined and the first prototype is in preparation.
1
2025Company founded
OÜ Reku starts in Pärnu County, Estonia, with the machine concept and the expert panel behind the thinning style.
2
First winter seasonPrototype in the field
A Usewood Combi with 3D vision and on-board AI, tested in planted birch on frozen ground. Remote operation first.
3
NextAutonomy
Tree selection and navigation move from the operator to the machine, trained on the data from every season.
4
Five yearsOwn fleet
Tens of REKU machines working in Estonia and the Baltics, then licensing the decision model to machine builders.
Markets
Estonia
Baltics
Finland & Scandinavia
Central Europe
Team
Foresters and engineers, one team.
Founder & CEOAndres Lainela
Educated in forest management. Leads REKU's strategy, partnerships and field work.
Co-founder & CTOLauri Leemet
Leads REKU's technology and machine development. Brings experience from building autonomous delivery robots in Estonia.
ML & vision advisorErle Leemet
Machine learning and computer vision specialist with a TalTech background. Advises on how the machine sees the stand.
Technical mentorTiit Kosenkranius
Creator of the Usewood machine, with more than 40 years in forestry. The source of our thinning knowledge.
We work with experienced foresters from state and private forest management, and with Estonian universities on research in machine vision and motion planning.
Contact
Together for a smarter forest.
Forest owners, research partners and investors: we would like to hear from you. Tell us about your stands, your questions or your plans.
Join us
Build the first machine with us
We are a small team. If you work in robotics, motion planning, machine vision or forestry and want to help build the prototype, write to us and tell us what you would work on.
- andres.lainela@gmail.com
- Company
- OÜ Reku · Reg. code 17213585
- Location
- Uulu Tehnopark
Uulu, Häädemeeste Parish
Pärnu County, Estonia - linkedin.com/company/reku-forest