RackKI

The FZI is working on robots that provide support with sensitive manipulation in this scenario. Robots can use human reference data to learn the interface-specific handling strategies for plugging in the cards. These force strategies are stored in a Deep Neural Network and optimized in a second simulative phase. Employing a targeted transfer learning phase, the AI system adapts the force-sensitive strategy according to the card used. The result is a new adaptive and force-based robotic solution.

In RackKI, an automated AI-supported solution was successfully developed. It implements a flexible handling approach with a terrestrial lightweight industrial robot and can plug in new cards intelligently and autonomously. A unique gripper and collision-free path planning were developed, enabling the cards to be transported from storage to the rack before the force-based plug-in.