This is a palletising robot: it picks boxes off the packing line conveyor and builds them into pallets ready for the warehouse. The standout requirement here isn’t speed or weight, it’s gentleness.
Marshmallow cartons are easily crushed or visibly deformed if gripped too firmly or stacked with too much placement force, a problem that doesn’t show up in a quick test run but becomes obvious once a few hundred pallets have gone through the line. The cartons also don’t always arrive in an identical position on the conveyor, so a fixed, pre-programmed pick point wasn’t reliable enough on its own. The customer needed a system that was fast enough to match the packing line, gentle enough to protect a delicate product, and flexible enough to handle more than one carton size or pallet pattern without a mechanical changeover.
A vision system tracks each carton on the conveyor and adjusts the robot’s approach for small variations in position or orientation, rather than assuming every carton arrives in exactly the same spot. The end-effector and motion profile were tuned for low-contact-force pickup and a soft, controlled placement onto the pallet, avoiding the sudden stops or hard contacts that would crush the product. Pallet pattern, carton size, and layer count are all selectable from the operator interface, so switching between product variants is a recipe change rather than a retooling job.
Pallets come out built to a consistent pattern and density every time, with none of the variability that comes from manual stacking under time pressure, and without the repetitive lifting that this task otherwise puts on staff. The vision-guided, low-force approach we used here is one we reuse anywhere a product is delicate enough that “just stack it like a box of bricks” isn’t good enough.