Vision-guided symmetric dual arm picking of large objects with human-like robot motions

Dual-arm robot picking of large objects in industrial settings, which is often performed alongside human operators, requires not only precise control of multiple arms for stable grasping but also a safe and trustworthy human-robot interaction. A possible way to fulfill the latter requirement is the implementation of human-like movements, which enhances motion predictability by workers. In this paper, we propose a technical integration framework combining passivity-based adaptive force-impedance control for multi-arm manipulation, a vision-based system and a human-like Cartesian motion planning algorithm for dual arm picking of large objects. Our approach was tested in experiments with real robots handling large boxes.