Mycionics: a decade of robotic mushroom harvesting, from farm-shack prototype to modular farm robots
Key points
Key takeaways from a YouTube presentation by Mycionics CEO Stefan Glibetic to the AgRobotics Working Group (July 2026):
Origin. In 2013 mushroom farmer Murray Good (president of Mushrooms Canada)approached the University of Western Ontario, where Glibetic was an undergraduate robotics student; a master's project yielded the first harvester prototype that could scan and pick mushrooms without damage by 2017, and the company was set up inside Good's farm in Putnam, Ontario.
The problem. Harvesting labour represents roughly 40–50% of mushroom production costs, and with millions of mushrooms a week to pick by hand, labour shortages, tightening immigration rules and an ageing workforce are making hand harvesting unsustainable across the industry.
First full-stack system. In 2022 Mycionics deployed robots that handled the entire harvest job — moving through the beds, finding, picking, cutting and packing — at two Ontario farms (Putnam and Hamilton); robotic picking proved more consistent than humans, with no damage over week-long cooler tracking.
The pivot. After a decade and over 20 generations of systems, the company stepped back in 2024: the full-stack robot's 35-motor complexity made it costly, hard to wash and prone to failure, so it modularised its core tech — soft-touch grippers, 3D vision and arms — into waterproof, industrially reliable modules designed to fit existing farm infrastructure.
Economics. The redesigned systems achieved roughly 3x the throughput at half the cost, cutting payback periods to commercially viable levels; for farmers the target is at least a 50% reduction in harvesting labour while matching human quality.
Yield data. Because the robots see every mushroom, Mycionics builds a digital twin of each farm; its crop-scouting and pointing system (green dots telling pickers which mushrooms to take) has averaged a 7.8% yield improvement over 90 days, and a 1mm larger cap diameter equals ~4% more yield — 17–18,000 lbs, or roughly $80–90k extra revenue per bed a year.
New generation. Theo latest systems, demoed at Mushroom Days inthe Netherlands, use four arms and a 75% robotic/ 25% human labour split, leaving delicate, higher-skill tasks to people; pushing toward 100% automation, Glibetic argues, makes the last 20% exponentially harder and uneconomic.
Data platform and roadmap. Farmers now get a dashboard with per-bed crop maps and microclimate data (where one sensor box currently covers 500–1,000 m² of mushrooms), plus tools to plan picking by mushroom size;; Mycionics wants to feed that ground-truth crop data back to genetics, irrigation and farm infrastructure — uplifting the dominant A15 strain, which hasn't been evolved in 15–20 years — and to support continuous mini-flush harvesting instead of the traditional 2–3 flushes.
Read more: YouTube · The AgRobotics Working Group