Investigating AR Assistance for Human-Robot Collaboration in Mould Assembly “in the Wild”

摘要

Augmented reality shows great promise for supporting human-robot collaboration. However, investigating this “in the wild” with participants from industry is mostly uncharted territory. Addressing this, we report on a field study evaluation of ARTHUR, an open-source authoring tool for augmented reality-assisted human-robot collaboration. To facilitate this domain-specific study, we extended ARTHUR with additional design components, improved the authoring experience, and support for collaborative authoring. The study took place at a local company, with 16 participants from three stakeholder groups: robot engineers, UX designers, and operators. The task was to assemble and disassemble a subset of an injection mould together with a collaborative robot, while supported by an AR-interface created in ARTHUR. The study revealed a strong desire of operators to be assisted in their daily tasks by robot(s) that serve as an extra (strong) set of hands, and augmented reality to guide the process and communicate robot intent. Engineers confirmed the feasibility of integrating ARTHUR in future semi-automated workflows and UX designers appreciated the authoring capabilities and proposed further features.