Skip to main content

RobVision

RobVision is RobCo's vision add-on for RobFlow. It pairs a 3D camera with trained pose estimation models so that a Robot can find out where an item actually is, instead of being taught a fixed Pose for it. The camera captures the scene, the models detect the known parts in it and compute a Pick Pose, and the Flow moves the Robot to that Pose.

The add-on brings its own hardware — a RobVision Device with a supported camera, mounted on the Robot or fixed next to it — and the software that runs the vision solutions on it. Once the device is registered under RobVision devices, RobFlow gains the RobVision view for setting up devices and solutions, the RobVision Trigger, RobVision Picking and RobVision Feedback Nodes, and the RobVision Widget for the Dashboard.

About this documentation​

This documentation provides a guide for configuring RobVision. It covers topics including camera setup, calibration, pose estimation, and the process of training and deploying models.
The system is divided into two main parts: RobVision and RobCo Studio.

  • RobCo Studio is used to train models that identify the location of objects.
  • RobVision executes vision solutions on the Robot for example to pick items. The software is integrated in RobFlow.

The following images show the steps to set up the Vision system.

The steps to set up the Vision system