Our paper, “CoVA: A Virtual Assistant for Virtual Reality Conferencing”, was presented at the Proceedings of the 21st EuroXR International Conference (EuroXR 2024).
Abstract
Teleoperation and collaboration are among the key pillars of business work, where services and demand are spread over a very wide economic market, such as the European Union. What characterizes their importance is not only remote exchanges, but also the ability to intervene or assist people without having to travel, as in the example of machine maintenance and repair. This is made possible by the integration of IoT and artificial intelligence into this vast technological field, which is further accentuated by extended reality.
CORTEX² project is in line with this vision to bridge the divide between widespread videoconferencing tools and state-of-the art XR-based solutions, democratizing the uptake of next-generation Extended Reality tele-cooperation among many industrial segments and SMEs.
One of the many promising developments in this space is the integration of AI-based conversational agents within XR environments (Reiners et al., 2021). When combined with XR applications, virtual agents can facilitate real-time collaboration, information retrieval, and task automation. However, this combination presents several challenges. In multi-party dialog contexts, where participants interact simultaneously, AI conversational agents must accurately handle real-time speech recognition and response generation (Clark et al., 2019). Achieving this at scale requires the use of highly efficient tools and models to minimize latency and ensure smooth, uninterrupted conversations, particularly when managing multiple users concurrently.
In the rest of this paper, we will illustrate the contribution of artificial intelligence to extended reality through an XR video conferencing application, where an AI virtual assistant plays a critical role in business meetings.
Authors
Alexis Lombard, Yazid Benazzouz, Galo Castillo López, Gaël de Chalendar, and Jean Pierre Lorré
Read the full publication
Access all our CORTEX2 publications.