Learning to Assist Bimanual Teleoperation using Interval Type-2 Polynomial Fuzzy Inference

Assisting humans in collaborative tasks is a promising application for robots, however effective assistance remains challenging. In this paper, we propose method providing intuitive robotic based on learning from human natural limb coordination. To encode coupling between multiple-limb motions, use novel interval type-2 (IT2) polynomial fuzzy inference modeling trajectory adaptation. The associated coefficients are estimated using modified recursive least-square with dynamic forgetting factor. We to employ Gaussian process produce robust motion predictions, and thus address the uncertainty measurement noise of system caused by interactive environments. Experimental results two types interaction demonstrate effectiveness approach, which achieves high accuracy predicting assistive enables perform bimanual only one limb.