The latest generation of industrial grippers works together with cameras, which provide them with 3D shots of the objects to be transported. A self-learning AI system processes the pictures and grips the parts, and constantly learns from this. Schunk’s SVH 5-finger hand , for which the manufacturer has deliberately limited the number of component variations, thereby streamlining the classification and training process, is one of the models used here.

At a demonstration of the system’s capabilities, manufacturer Schunk interlocked plastic building blocks in several combinations and instructed the gripper to take them away. After just a few learning cycles, the system had adapted its algorithms and increased the secureness of the grip. This enabled it independently to classify the combinations and arrangements of components and increasingly rely on its learned empirical values while working.