Closing the Loop: Data-Driven Control for Pneumatic Grippers

Chef robots pick food thousands of times per hour using pneumatic grippers, but gripper opening speed varies across robots and degrades over time, making manual tuning unreliable. Chef Robotics built a layered system combining a flowmeter, regression modeling, a neural network, and online adaptive control to measure gripper behavior and automatically correct for it without manual intervention.