Skip to content

HomeProjects › 3D bin picking cell,
vision guided

Robotic cell Catalogue solution Mechanical · Electrical · Automation · Vision In-house development

3D bin picking cell,
vision guided

Feeding a production machine with parts taken from bulk, without a vibratory bowl feeder and without mechanical adjustment at every change of reference. A robot, a 3D camera and a program that decides — developed in house, available from our catalogue.

Equipment
Robotic feeding cell
Function
3D bin picking and machine loading
Robot
ABB
Vision
Keyence, 3D recognition
PLC
Siemens
Origin
SIP in-house development

The problem with bulk

Feeding a machine with parts from a bin is one of the least rewarding jobs in a workshop. The classic answer is a vibratory bowl feeder: it works, but it is noisy, it wears, it needs maintenance, and above all it is built for one single reference. Changing part means changing bowl, or a long readjustment.

The second classic answer, 2D vision, struggles with parts that are stacked, shiny, dark or variable in colour.

Our answer

A cell where a 3D camera reconstructs the relief of the bin, identifies which parts can be gripped and tells the robot which pick to make. The program handles the awkward cases: unreachable part, uncertain grip, nearly empty bin.

  • 3D rather than 2D recognition: the measurement is of relief, so neither colour, nor surface finish, nor gloss disturbs detection.
  • Wide range of parts: changing reference is a matter of parameters, not of tooling.
  • Case handling built into the program: every difficult pick situation has a planned response, rather than a fault stop.
  • Turning over the last parts: a dedicated mechanism repositions parts left unusable at the bottom of the bin, so the bin is emptied rather than the cell stopping at ninety per cent.
  • Simulation before integration: trajectories, cycle time and success rate proven offline, on a digital mock-up, before the first real part.

What it changes

No bowl feeder means no mechanical wear parts in constant friction: maintenance drops and the noise disappears from the workstation. Throughput is high because the robot does not stop for an edge case, and because the bin is emptied completely.

This is not a study. The cell exists, it runs on our premises, and its software core is part of our solution catalogue: what remains on a client project is adapting it to their parts, their machine and their throughput.

Bin picking simulation environment: 3D view of the robot above a bin of parts, picked object counter, gripper and 3D recognition settings.
The bin picking simulation environment. In the centre, the robot and the bin reconstructed in 3D, with the calculated pick trajectory. Above, the object counter and the success rate for the run. On the right, the gripper settings, the import of the part's CAD model and the bin parameters. The parts used for this demonstration have been blurred.
Advantages

Why 3D bin picking rather than a bowl feeder

One machine, all your parts

A bowl feeder is designed for one reference. Here, changing part is a matter of parameters and a CAD model, not tooling.

Indifferent to appearance

Recognition works on relief. A black, shiny part, or one that varies in shade between batches, is detected in exactly the same way.

Little maintenance

Without a bowl, there are no parts in constant friction and no mechanical settings that drift. And the noise leaves the workstation.

The bin emptied to the bottom

The turning mechanism repositions the last unusable parts, instead of letting the cell stop with a residue in the bin.

Planned cases, not faults

Uncertain grip, unreachable part, nearly empty bin: each situation has a programmed response. That is where throughput comes from.

A proven core

The solution exists and runs. On a project, the effort goes into your parts, your machine and your throughput.

A loading station to automate?

A bowl feeder to replace, bulk parts to sort?

Send us the part you want to pick from bulk and the throughput you are aiming for. We will tell you honestly whether 3D bin picking is the right answer — and if it is not, which one is.