AI for the forestry and wood-products industry in NWO

From harvest scheduling to lumber grading, machine learning is entering the forestry value chain. Here is where the technology stands, what is confirmed in Canadian operations, and where a Northwestern Ontario mill or woodlands company can start.

AI is entering the forestry and wood-products value chain at three practical points: harvest planning and woodlands scheduling, where machine learning tools analyse historical field data to reduce unplanned operational days; mill-floor machine vision for lumber grading, where deep learning systems grade wood from all four sides in a single pass faster and more consistently than rules-based scanners; and forest inventory, where Ontario has invested $84.5 million in a LiDAR-based modernization program replacing aerial photo interpretation across Crown forest management units, including the White River Forest Management Unit in Northwestern Ontario. Northwestern Ontario's forestry sector supports approximately 2,700 direct jobs and represents 60 per cent of Ontario's total provincial forestry employment, according to FedNor's Northern Ontario Economic Overview. FedNor's Regional Artificial Intelligence Initiative (RAII) explicitly lists forestry as a target sector for AI adoption funding in Northern Ontario — making the same program available to a forestry or wood-products business as to a mining or manufacturing operation.

Harvest planning: what AI optimization tools actually do

Harvest planning is where AI has the clearest, best-documented track record in Canadian forestry. The core problem is data volume and variability: woodlands operations involve dozens of cutblocks, multiple contractors, equipment spread across hundreds of kilometres of road, and weather that changes conditions daily. Machine learning tools analyse historical field data — volumes hauled, contractor productivity, road conditions, seasonal patterns — to produce forecasts more accurate than rules-based scheduling.

Remsoft, a Canadian forestry software company, published analysis in Wood Business (Canadian Forest Industries) showing that a 5–10 per cent improvement in harvest volume and productivity prediction accuracy translates to more than 1,000 fewer unplanned operational days per year and 600,000 cubic metres of scheduling noise removed from a six-million-cubic-metre harvest program. This is company-published analysis, not an independently audited figure, but the operational mechanic it describes — how small accuracy gains compound over a large program — is concrete and grounded in operational data. In November 2025, IBM and Polytechnique Montreal announced a collaboration through IBM's Impact Accelerator to build AI and quantum-enabled decision-support tools specifically for Canada's forest supply chain, combining machine learning, digital twins, and multi-objective optimization for harvest planning and yield forecasting.

Mill floor: AI lumber grading

At the planer-mill level, AI enters through machine vision grading. Traditional grading uses rules-based scanners with fixed measurement thresholds. AI-based systems — such as Lucidyne's Perceptive Sight platform and USNR's Deep Neural Network (DNN) grading — run deep learning models trained on large image datasets to classify lumber from all four sides in a single pass, detecting knots, splits, worm holes, bark pockets, shake, and decay more consistently at high throughput speeds. Wood Business (Canadian Forest Industries) reports Lucidyne's system is in operation at more than 65 planer mills across North America and Australia, and USNR reports well over 100 systems deployed on its DNN grading line. No specific Ontario mill deployments are confirmed in publicly available sources — sawmill technology investments are not routinely disclosed publicly — but the equipment is sold and serviced in Canada. For an NWO sawmill or planer mill planning a scanner or optimizer capital refresh, AI-capable grading systems are now within the scope of that decision rather than a separate AI initiative.

Forest inventory: Ontario's LiDAR program and what it means for woodlands operations

Ontario's Ministry of Natural Resources and Forestry has invested $84.5 million in a Forest Resources Inventory (FRI) modernization program covering approximately 555,000 km² of Crown forest land and wetlands. The program replaces aerial photo interpretation with LiDAR — aerial laser scanning that generates three-dimensional point clouds from which wood volume, tree height, and basal area are estimated using predictive modelling. As of Ontario's 2025 Forest Sector Strategy Progress Report, 415,000 km² of LiDAR data had been collected toward a 455,000 km² managed-forest target. The White River Forest Management Unit, in Northwestern Ontario, received its first LiDAR-based FRI in 2024–25.

For an NWO woodlands operation, this matters practically: harvest plans, sustained-yield calculations, and environmental assessments increasingly draw on LiDAR-based inventory data rather than older aerial photo estimates. Operators working with Crown timber allocations in units where LiDAR data is now current have a more accurate and current data layer — which is also the input layer that AI harvest optimization tools depend on. The inventory modernization program is administered by the province and applies to Crown forest management units; it does not require action from individual operators.

Where an NWO forestry or wood-products business can start

The accessible entry points are the ones that work on existing data and existing infrastructure:

• Harvest scheduling tools: Cloud-based optimization platforms such as Remsoft's are available by subscription to woodlands operations of varying scale. The prerequisite is having harvest and contractor performance data in structured, queryable form — data organization comes before vendor evaluation.
• Predictive maintenance: Sensor-based monitoring on heavy equipment — processors, skidders, haul trucks — feeds operational data to diagnostic tools that flag maintenance needs before failures occur. Several equipment OEMs bundle this with equipment purchases; third-party industrial monitoring tools are available independently. The return is clearest for operators running fleets far from service facilities, which describes most NWO woodlands operations.
• Grading and scanning capital decisions: For sawmills and planer mills scheduling a scanner or optimizer refresh, AI-capable grading systems from vendors such as Lucidyne and USNR are now within the scope of that capital decision. Confirm equipment eligibility under FedNor RAII before assuming it qualifies — a FedNor officer contact is required before application.

FedNor's Regional Artificial Intelligence Initiative (RAII) lists forestry explicitly as a target sector. For for-profit organizations in Northern Ontario, the program covers up to 50 per cent of capital costs (repayable) and up to 75 per cent of non-capital costs (repayable). Confirm current eligibility and intake status directly with FedNor (1-877-333-6673) before applying — retail businesses are excluded, and a FedNor officer contact is required before a formal application is submitted.

Sources: FedNor RAII — forestry listed as target sector, funding terms (verified directly): fednor.canada.ca/en/our-programs/regional-artificial-intelligence-initiative-raii-northern-ontario | FedNor Northern Ontario Economic Overview — approximately 2,700 direct forestry jobs in Northern Ontario, 60% of provincial forestry employment (2024 data): fednor.canada.ca/en/resources-and-tools/northern-ontario-economic-overview | Ontario Forest Sector Strategy 2025 Progress Report — $5.4B GDP contribution, 128,000+ jobs, $8B exports, $84.5M FRI investment, 415,000 km² LiDAR collected, White River FMU LiDAR delivery 2024–25, FSIIP $72M / 22 projects / 358 new jobs: ontario.ca/page/success-glance-ontarios-forest-sector-strategy-2025-progress-report | Ontario Forest Resources Inventory — LiDAR program: ontario.ca/page/forest-resources-inventory | Northwestern Ontario sawmill operations (3 mills — Atikokan, Ignace, Thunder Bay; 514 million board feet annual capacity; 467 employees; $177.2M economic impact): domtar.com/our-location/northwestern-ontario-operations/ | Remsoft — AI harvest planning, 5–10% accuracy improvement, 1,000+ days saved and 600,000 m³ noise removed on a 6M m³ program: woodbusiness.ca/harvest-accuracy-and-ai/ | Wood Business — AI lumber grading case study, Lucidyne Perceptive Sight and USNR DNN grading (65+ North American deployments, 100+ systems): woodbusiness.ca/ai-in-action-a-case-study-on-intelligent-lumber-grading/ | IBM and Polytechnique Montreal — AI for Canadian forest supply chain (November 2025): canada.newsroom.ibm.com/2025-11-13-IBM-and-Polytechnique-Montreal-Launch-AI-Initiative-to-Strengthen-Forestry-Supply-Chain | Nextfor — NWO forest innovation working group: nextfor.ca/focus-areas/forest-innovation/ | CRIBE, Thunder Bay: cribe.ca