Skip to content

Goods2Robot: optimised packing with a cobot

Packing a box by hand is simple. Packing it so that less than half of it is air, order after order, is not.

That second part is becoming a requirement. The EU packaging regulation, PPWR, includes a requirement that the share of empty space in packaging stays below 50 %. For anyone who picks and packs customer orders with many different products, every box becomes a small puzzle, solved again for each order.

Goods2Robot solution with optimised packaging: a cobot packs products from a conveyor into a crate

How Goods2Robot works

Goods2Robot is a goods-to-robot cell with an ABB GoFa cobot. The goods are fed to the robot, which picks and packs them per customer order. The cell uses product metadata, or measures each product dynamically, and plans where each item goes in the box or crate.

AI where it helps the robot see

Two different jobs happen in the cell, and we have made different choices for them.

We use AI where it helps the robot see, in our own vision system, which does not need massive training. That is where AI does real work in Goods2Robot.

The packing itself, our 3D Tetris, is fully deterministic. We tested AI for this kind of 3D Tetris optimisation too, but it was too sensitive to the training data. So we built a packing solution that uses no AI and needs no training.

What the fill rate looks like

Fill rate depends on what is in the order. Two examples from our presentation: an order of 20 items filled 86 % of the volume, and an order of 19 items filled 85 %. In both, the empty space was well below 50 %.

Every pick is recorded, so there is full track and trace of what went into each box. The manual picking and packing work is taken over by the robot.

Where it fits

Goods2Robot is built for operations that pack mixed customer orders in boxes or crates. For larger volumes and whole pallets, the same deterministic 3D Tetris builds mixed pallets and parcel pallets. You can read more in Mixed pallets and parcels with full traceability.

Since the result depends on your products, we would rather start from your situation than from a promise. Tell us how you pack today, and if you can, send a sample of your order data. Then we can see together what a cell could do with it. Contact us.