A Framework for Composite Layup Skill Learning and Generalizing Through Teleoperation.

Area of Science:

• Robotics
• Human-Robot Interaction
• Composite Materials Manufacturing

Background:

• Efficient and intuitive human-robot skill transfer remains a challenge, particularly in complex manipulation tasks like composite layup.
• Existing teleoperation methods often lack the necessary intuitiveness and efficiency for seamless skill acquisition by robots.

Purpose of the Study:

• To develop and investigate an impedance control-based framework for effective human-robot skill transfer in composite layup.
• To enable intuitive skill acquisition through teleoperation and facilitate human-robot collaboration.

Main Methods:

• An impedance-based control architecture in task space was developed for telemanipulation.
• Variable impedance control was utilized for compliant robot-environment interaction and smooth transitions.
• Dynamic Movement Primitives (DMP) based Learning from Demonstration (LfD) was employed to model human manipulation skills.

Main Results:

• The framework successfully enabled human-robot skill transfer for composite layup tasks.
• The learned skills demonstrated generalization capabilities across different component shapes and orientations.
• The system facilitated compliant interaction and smooth transitions during the teleoperation process.

Conclusions:

• The proposed impedance control framework offers an efficient and intuitive method for human-robot skill transfer in composite layup.
• The approach supports human-robot collaboration through teleoperation, enhancing manufacturing processes.
• The generalization of learned skills across varying task parameters shows the robustness of the developed system.