Editor's note
Four stories this week, and not one new machine.
It is July, the Nordic industry is on holiday, and the news flow thins out accordingly. What is left is oddly consistent. Sensors bolted onto fans that already spin. Optimisation laid over a wood flow that already runs. Tree species read out of satellite images that were already being collected. Tree height and species from an ordinary drone camera, no LiDAR required.
Nobody bought anything heavy. Every one of these projects makes better use of something that was already there, and none of them needed a purchase order with a comma in it. That is a cheaper kind of progress than the industry usually plans for, and it is easy to miss precisely because there is nothing to stand next to in a photo.
Axel
Metsä Group puts Qutwo's AI-and-quantum platform to work on wood procurement
Metsä Group will run wood routing from forest to mill, and its tissue converting lines, on Qutwo OS, a platform from the Finnish company Qutwo that combines AI with quantum computing. Metsä already uses AI to forecast forest damage and price timber deals.
SCA bolts wireless sensors onto the roof fans at its Tunadal planing mill
SCA Wood Scandinavia is fitting the process-critical fans at Tunadal with wireless vibration and temperature sensors from SPM Instrument. The fans sit on the roof, hard to reach and easy to ignore until they seize or overheat. The point is to catch the failure early: plan the maintenance stop instead of taking it, and keep a hot bearing from becoming a fire.
Lund researcher maps which trees grow where in Götaland, and gives the maps away
Abdulhakim Abdi at Lund University has built the first large-scale classified map of dominant tree species in Götaland, the southern third of Sweden where oak and beech break up the spruce-pine-birch monotony of the north. It is made from satellite data and machine learning, published openly at treespecies.se, and every prediction carries an uncertainty estimate.
Mila releases an open benchmark for reading tree height and species off a single drone photo
Researchers at Mila, McGill and Université de Montréal have published BIRCH-Trees, a benchmark for estimating the height and species of individual trees from ordinary RGB drone imagery, spanning temperate forest, tropical forest and boreal plantation. Their model beats a foundation-model baseline on height while using roughly half the parameters. Code, weights and data are open.