This robot system picks individual items from a jumbled, continuously moving stream of objects. It uses a 3D camera to identify objects, then a robot with a gripper (made of several parts) adjusts its grip to match the chosen object's shape. The robot picks the object and places it neatly onto a conveyor belt leading to a sorter, or directly onto a sorter. The system also learns to improve its picking accuracy by comparing images of objects before and after they are placed.
Why it matters: Filed before significant advancements in real-time machine learning for complex object recognition and robotic manipulation. Training and deploying robust models for dynamic bulk handling is now easier.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI