TOMRA Launches AI Potato Sorter for Growers
The TOMRA 4A uses LUCAi AI and optical inspection to sort potatoes at up to 100 tonnes per hour.
Updated on Nov. 5, 2026 in Artificial Intelligence
TOMRA Food has launched the TOMRA 4A, an AI-powered optical potato sorter that removes foreign material such as stones and mud clods with 92 percent accuracy. The machine launches on November 5 for unwashed potatoes and will debut at Interpom 2026 in Kortrijk, Belgium.
Why it matters
Growers face shorter, weather-dependent harvest windows, labour shortages in seasonal work, and rising growing and storage costs. The TOMRA 4A is designed to increase the speed, capacity, and accuracy of sorting in response to these pressures.
The TOMRA 4A combines dual sided Pulsed LED+ inspection, precise ejection technology, and the LUCAi AI platform, with three-way sorting into good potatoes, out-of-spec potatoes, and foreign material streams. Pre-trained models are ready from day one, and machine learning lets operators teach the system to classify crop-specific defects, though independent validation of these figures is not reported.
The players
TOMRA Food
TOMRA Food is the food-processing division of TOMRA, a Norwegian company that has developed sensor-based sorting technology for decades.
The details
The machine uses a single XCD user interface so one operator can manage the entire line. TOMRA says the platform is being developed for multiple applications, with onion sorting set to launch on all TOMRA 4A machines in time for the next harvest season.
Timeline
November 5: TOMRA 4A launches for unwashed potatoes.
November 29 to December 1: TOMRA 4A debuts at Interpom 2026 in Kortrijk, Belgium.
Next harvest season: Onion sorting launches on all TOMRA 4A machines.
Roadmap
The TOMRA 4A extends TOMRA's roughly 14 years of optical potato sorter development by layering AI classification onto established sensor hardware. The move follows a broader pattern in food processing where automated inspection is being tied to adaptive machine learning rather than fixed rule-based sorting.
Potato growers using the TOMRA 4A could process larger volumes with fewer operators, as the XCD interface lets one person run the full line. Pre-trained models mean a line can start sorting from day one without extended setup, though the source does not give pricing.
The takeaway
Automated optical sorting with adaptive AI is becoming a practical response to labour shortages and tighter harvest windows in potato production. Growers evaluating new sorting equipment may want to ask how quickly a system can be retrained for their specific defects and crop conditions.
What happens next
Onion sorting will launch on all TOMRA 4A machines in time for the next harvest season, and the machine debuts at Interpom 2026 from November 29 to December 1.
Further reading
Explore more Artificial Intelligence coverage to see how machine learning is being applied across industries.
