Industries
/
Manufacturing

Manufacturing robotics software

Software for robotic welding, painting and assembly, and for upgrading robot cells that are already in production.

Get In Touch
Robotic welding arm with 3D vision

What we build for manufacturing

Robotic welding with 3D vision

A camera on the robot finds the part, so each weld follows the real part instead of a taught path.

  • Depth camera on the wrist locates the workpiece
  • Weld paths set from perception, with no manual teaching
  • Tested in Isaac Sim before the real cell

Robotic painting and coating

Spray paths for parts that change in shape or position.

  • Spray paths planned from the part's 3D shape
  • Added to the paint robots you already have
  • Coverage checked in simulation first

Vision to locate parts on robotic assembly lines

Cameras and 3D sensors that find each part, so the robot picks and places it correctly.

  • 3D scans matched to the part's CAD model
  • Parts picked even when they are not in a fixed position
  • Training images generated in simulation when real data is short

Upgrades for robot cells already in production

Perception and adaptive motion planning added to welding, painting and palletizing cells you already run.

  • Starts with a review of the cell
  • Sensors added only if they are needed
  • Usually done within weeks

FAQs

+
How does vision-guided robotic welding work?

A depth camera on the robot's wrist scans the part, the software matches the scan to the part's CAD model, and the weld path is set from where the part actually sits. Black Coffee Robotics builds this software, so a part placed slightly differently does not mean teaching the robot again.

+
Can vision be added to a welding or painting robot that is already in production?

Yes. Black Coffee Robotics adds perception and adaptive motion planning to welding, painting and palletizing cells already in production, and the existing robots and controllers stay in place. An upgrade starts with a review of the cell, sensors are added only if needed, and it is usually done within weeks.

+
Can a vision model find parts when there are few real images to train on?

Yes. Black Coffee Robotics generates synthetic training images in simulation, with random colours, textures and lighting, so a vision model can learn to find parts on an assembly line before much real data exists.

Want to reduce costs and time to market for your autonomous robots?