Glossary

Bin picking

Bin picking is a robotic task in which a robot, guided by a vision system, locates and grasps individual parts from a container where they lie in random positions and orientations.

A bin picking cell has three parts. A vision system, usually a 3D camera or structured-light scanner, captures the bin. Software identifies candidate parts, estimates their position and orientation, and chooses one that can be reached and gripped without collision. The robot then picks it with a suction cup, parallel gripper, magnet or custom tool and places it, often into a fixture or a machine.

The difficulty depends on how structured the bin is. Parts laid out in known positions are simple. Randomly piled parts are hard: they overlap, hide each other's features, tangle, reflect light or sit against the bin wall where the gripper cannot reach. Shiny metal and dark or transparent parts are common problems for cameras. Many systems re-image after each pick and need a strategy for the last parts in the corner.

Classic approaches matched CAD models to the 3D image. Machine learning now helps with grasp selection and with mixed or unknown items, a problem pushed forward by warehouse picking competitions. For a given part, the key numbers are pick success rate and cycle time, which should be measured on real parts and real bins, not demo samples.

Questions about Bin picking

Why is picking from a random pile so hard?

Because the parts overlap, hide one another's features, tangle, reflect light or rest against the bin wall where the gripper cannot reach them. Shiny metal and dark or transparent parts also cause trouble for cameras.

What does a bin picking cell consist of?

Three parts. A vision system, usually a 3D camera or structured-light scanner, images the bin. Software finds candidate parts, estimates their position and orientation and selects one it can grip without collision. A robot then picks it with a suction cup, parallel gripper, magnet or custom tool.

When is bin picking not worth it?

When the pick rate is so low that an operator is constantly clearing misses. The numbers that matter are pick success rate and cycle time, and they should be measured on real parts in real bins, not on demo samples.

Bin picking in real plants

  • MAB Möbel and MVTec Software: At MAB Möbel in Muotathal, a HOMAG robot cell uses a 3D laser scanner and MVTec HALCON to pick wooden workpieces from a chaotic stack and load a CNC machining center. production
  • FANCL and Mujin: At its Osaka logistics center, FANCL uses MujinOS-powered piece-picking robots with 3D vision for mixed-SKU picking, sorting and six-side barcode scanning. production
  • Logisteed and Mujin: LOGISTEED uses MujinOS-powered piece-picking robots with 3D vision and barcode verification for the secondary sorting of pharmaceutical products. production
  • Chroff Kunststofftechnik and Yaskawa Electric: Chroff Kunststofftechnik sorts randomly placed injection-molded parts into containers with an FMB handling system: a Yaskawa GP12 robot guided by a Keyence AI camera. production

All deployments

Vendors linked to Bin picking

Mujin (2) · MVTec Software (1) · Yaskawa Electric (1) · KUKA

From our tracker and company profiles, not a full market list · In brackets: deployments tracked · All companies · No paid placement

Related terms

Sources

Checked September 30, 2026 · Updated October 3, 2026 · All terms · Spotted an error? Tell us.

Monday digest

See these terms in the news, weekly.

8 to 12 sourced items · every Monday morning · free