Vision 2021: System for reliable object recognition

The system adapts to the respective conditions by using the neural networks and also reliably detects complex and reflective surfaces such as metal and calculates possible grip points. Via the integrated PLC communication, the system supplies the robot controller with collision-free waypoints and thus enables handling in non-sorted boxes and for complex workpieces that were previously difficult to automate. The light field-based multi-camera approach also results in fewer obscurations and a high depth of field is retained even with increasing distance.

The associated simple and intuitive software supports the user in adapting the system individually for individual tasks and ensures commissioning in the shortest possible time. A password-protected manual mode enables parts to be drawn in manually during the process and the neural networks to be exchanged during operation.