AIToday
RoboticsRobotics & Automation NewsPublished: Oct 9, 2026, 04:00 JST

Stäubli robots and QING's STAQ grade 100,000 peaches an hour in Greece

Stäubli robots and QING's STAQ grade 100,000 peaches an hour in Greece

3 Key Points

  1. What happened

    At Agrophoenix's Greek plant, eight Stäubli SCARA robots run QING's AI- and vision-based STAQ system, grading 15 tons of peaches per hour — about 100,000 peaches, or 200,000 halves.

  2. Why it matters

    The two-season effort hinged not on AI or software but on the gripper material that touches slippery peeled peaches, showing that food-grade handling, not the algorithms, slowed the system down.

WHO IT HITSFood-processing plant managers and quality-control teams at fruit canners and packers may see that AI vision plus washdown-capable robots can replace manual peach grading at high speed. Robot and gripper suppliers in food handling will face pressure to prove hygienic compliance and gripper reliability, not just AI accuracy.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Agrophoenix was founded in 2018 in Greece's Imathia region with the stated aim of differentiating itself from standard fruit-processing models. That ambition shows in its production choices: a new line for fruit balls in pouches, and aseptic diced fruit packed in 200-liter bins usable year-round for cup products. The peach-grading line was built with a specific goal — removing fruit with pit fragments. After washing, steam-peeling, cutting into halves, and depitting, the fruit is pre-inspected by a vision system, mainly for color, before reaching the eight conveyors.

QING, a Dutch startup based in Arnhem, developed the STAQ framework — See, Think, Act — for the food industry. QING's commercial director Will Uijting says the same software and functions work across products like meat and biscuits, with grippers and algorithms adapted per product. The robots are Stäubli TS2-80 units in the HE version, which have washdown capabilities: protected joints, corrosion-resistant materials, and inner pressure that keeps vapor, steam, or water out during cleaning. The cells require frequent cleaning — every four hours the lines stop for 20 minutes for automatic CIP-like cleaning, and each Sunday a more intense cleaning with food-grade detergent takes place.

This is not a laboratory demonstration but a completed first harvest season, which suggests the approach is moving from pilot to routine production. The next engineering target is harder: an automated system to remove pit fragments from peach halves, before or after sorting, that Ioannidis describes as the next big thing.

FAQ
How many peaches per hour can the system handle?
During the season the eight STAQ cells process 15 tons of peaches per hour, which Agrophoenix calculates as approximately 100,000 peaches or 200,000 halves. Around 3,600 halves can be rejected in that time.
What happens to peaches with pit fragments?
The system currently looks for pits and picks them into two channels: those with bigger pits go to automated re-pitting, while those with smaller pits go to manual inspection. Both are then fed back into the STAQ stations.
Why did it take two seasons to get the cells production-ready?
The hardest part was not the AI, software, or robots but the gripper and the material that contacts the peaches. Engineers had to find a conveyor material that holds the slippery peeled peaches and the right vacuum gripper to grab and release them.
Robotics & Automation NewsRead Original Article

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleHelm.ai signs $70 million in physical-AI model contracts