Preliminary investigation of augmented intelligence for remote assistance using a wearable display
Wei Ling Koh, Joyce Kaliappan, Mark Rice, Keng-Teck Ma, Hong Huei Tay, Wah Pheow Tan · 2017
Advances in augmented intelligence (AmI) and machine learning offers the potential to support aspects of remote assistance. This includes situations where a field operator requires the technical support of an expert, such as for assembly, maintenance or repair tasks. While current remote systems commonly rely on human intervention, we propose looking at the feasibility of an artificial agent to take on this role. Specifically, in this paper we present the findings of a remote assistance platform that allows an operator to receive text and video prompts via an AmI expert. A preliminary user study with a human researcher simulating the expert was conducted to compare three levels of AmI (low, medium and high), while participants wore a pair of smart glasses. Overall, statistical differences were identified in the number of prompts used. Further, we identified that participants' interaction with the AmI differed according to the level of perceived intelligence. From these findings, design recommendations include having segmented videos, active AmI monitoring and intervention, and more specific task prompts.