Ask most people how warehouse robots identify items, and they assume barcode scanning. Scan the label, confirm the SKU, execute the pick. That logic works until the label is damaged, the item is bundled, or a new product arrives before the system catalog is updated.

Barcode-based automation has a hard dependency: the item has to be cataloged before the robot can handle it. For operations with stable, predictable SKU sets, that works. For 3PL providers, e-commerce fulfillment, and anyone managing a rotating product mix, it creates a ceiling.

AI vision removes that ceiling.

CMES AI Robotics uses proprietary 3D vision technology that identifies items by shape, size, packaging type, and spatial orientation — not by a label. The system recognizes over 10,000 item types immediately and handles items it has never seen before by generalizing from what it knows about geometry, surface properties, and grip mechanics.

The practical result: a new SKU enters the facility, and the robotic piece picking system handles it without a reprogramming cycle. Bundled products, bubble wrap packaging, transparent packaging, reflective surfaces — the robot picks them. At 650 picks per hour. At 99.9% accuracy.

Barcode scanning is still part of the CMES system — for order verification. But the picking decision doesn’t depend on it. That’s the difference.

For operations evaluating warehouse automation, the question to ask any vendor is simple: what happens when the system encounters an item it hasn’t seen before? The answer tells you a lot about whether the technology is ready for a real production floor.

Visit www.piecepicking.com.