ActiveAR: Augmented Reality Task Support System With Proactive Context and Virtual Content Management.

Area of Science:

• Computer Science
• Human-Computer Interaction
• Artificial Intelligence

Background:

• Augmented Reality (AR) systems aim to enhance user efficiency but are often hindered by virtual content interfering with real-world tasks.
• Existing AR applications primarily focus on minimizing virtual content interference, rather than actively utilizing it for task guidance.

Purpose of the Study:

• To develop an innovative AR Task Support System that proactively guides users through tasks using context-aware virtual content.
• To overcome the limitations of current AR systems in uncontrolled environments and reduce development costs.

Main Methods:

• The system proactively searches for and tracks key objects in the real-world scene.
• Contextual information from tracked objects is used to automatically select and position relevant virtual content.
• Open-world prompt-based visual models are employed for few-shot or zero-shot object recognition, including uncommon items.

Main Results:

• The proposed AR Task Support System effectively guides users by actively displaying contextually relevant virtual content.
• The system demonstrates successful identification and utilization of both common and uncommon objects in real-world scenarios.
• Experimental results show significant advantages over traditional virtual content management systems in realistic usage situations.

Conclusions:

• The developed AR Task Support System enhances user efficiency by intelligently integrating virtual content with real-world tasks.
• The system's ability to handle diverse objects and environments broadens the applicability of AR beyond industrial settings.
• This approach offers a cost-effective solution for developing advanced AR task support applications.