Skip to Content

Other machines

Special machines we have designed for one job and one production line. Robots, vision and control, built around the customer's process.

  • Glass industry
  • AI vision

Vision-guided pick and place for glassware

Two high-speed Mitsubishi SCARA robots pick glasses from the cooling-oven conveyor and place them in order. A deep-learning vision system finds every glass, even on a moving mesh belt.

Speed
60 glasses per minute
Tracking
30 glasses at once
Missed glasses
Under 1%
Statistics
Every glass counted
Two SCARA robots above a conveyor, with a camera bridge and safety fencing
Two SCARA robots work side by side above the conveyor, behind safety fencing.

How it works

The robots and the vision system talk in real time, so each robot knows where every glass is as the belt moves.

  1. Glasses arriveGlasses leave the cooling oven upside down, spread across the belt.
  2. The camera finds themDeep-learning AI recognises every glass and tracks up to 30 at the same time.
  3. The robots pickVacuum grippers pick one or several glasses at once, synchronised with the moving belt.
  4. Placed and countedGlasses are placed on the outfeed conveyor in order, and every glass is counted for statistics.

See it on the line

Three vacuum grippers on one plate pick glasses straight off the oven belt.

Three vacuum grippers picking glasses from a mesh conveyor

New product? Teach the camera

The vision system learns a new glass from a small set of photos. No reprogramming of the robots is needed.

  1. Camera image of a glass on the mesh belt
    Take photosSave images of the new glass as it passes under the camera.
  2. The glass outlined and labelled in the training software
    Train the modelOutline the glass in a few images and let the software train the deep-learning model.
  3. Live detection with every glass marked on the belt
    Run it liveLoad the model into the camera. Every glass is found and its position sent to the robots.

Specifications

Typical configuration. Each system is adapted to the customer's products and line.

Picking speed60 glasses per minute
Glasses tracked at once30
Missed glassesUnder 1%, with correct setup
Glass diameter25 to 100 mm
Glass height20 to 320 mm
Glass weightUp to 2 kg
StatisticsCounting, ready for SCADA connection
Robots2 × Mitsubishi Electric RH-12FRH SCARA, 4-axis
Robot payload and reach12 kg, 850 mm
Robot speed and repeatability11 350 mm/s, ±0.015 mm
GrippersSchmalz vacuum, up to 6 per plate
VisionZebra Aurora Deep Learning AI, HIKROBOT cameras, optics and lighting
ControlMELSEC iQ-F PLC, GOT2000 touch panel
SafetySICK sensors and safety system, steel fencing
Two-robot glass orientation system mounted over a cooling-oven conveyor

Built around your line

The system is adapted to the space at the end of the cooling oven and to your products.

  • One or two robotsSized for the output of your line.
  • Defect inspection optionAurora AI can also detect deformations and defects on the glasses.
  • Easy accessFencing designed so operators can reach the oven conveyor quickly in an emergency.
  • TrainingOperator and maintenance training to fit your team.

Have a task no standard machine can do?

Describe your product, your line and what needs to happen. We will design the machine around it.