A Full Body Avatar-Based Telepresence System for Dissimilar Spaces

Abstract

Figure 1: Our telepresence system allows for different room environments by retargeting the user's placement and deictic gesture to the remote avatar so as to preserve the user's context of placement and interaction. In this figure, a user in the left space is sitting facing a display and pointing at a point in the display. The user is represented in the right space by a male avatar sitting on the right side of a display, pointing at the corresponding point in the display. A user sitting in front of the display in the right space is represented by a female avatar sitting to the left of the display in the left space. Small figures are user's egocentric views.

Partially Blended Realities: Aligning Dissimilar Spaces for Distributed Mixed Reality Meetings

“…To cope with room disparities, researchers have developed prototype systems that demonstrate various approaches to aligning distributed spaces. Solutions involve discretizing environments into functional spaces [45], warping the avatar's deictic gestures based on landmarks [44], adapting avatar movements [5,14,43]), and computing an optimal partial alignment [21]. We revisit these solutions in more detail in section 3.…”

“…Studies have shown that passive haptics supports a better sense of presence in the virtual environment [13,37], which can be used in systems for aligning physical and virtual environments [11,37,41]. Moreover, providing landmarks in the physical environment can enhance the ability to spatially reference shared objects [25,44].…”

Online Avatar Motion Adaptation to Morphologically‐similar Spaces

“…To retarget a user's deictic gestures to an avatar, gesture retargeting approaches have been proposed [KL20,YYCL21]. The user's head direction and arm posture are adjusted to make the avatar point to the same position toward the remote object as that toward the corresponding local object.…”

“…For example, several approaches deal with the modification of the user root path during locomotion without supporting the user's interaction with objects [CWS18,TSDS19] or only support human-object interaction motions within a pre-defined area around an object [KPBL16]. Other approaches propose placement retargeting methods for an avatar [YYK * 20] with upper-body gestures only [KL20,YYCL21]. Despite their reported successes, these approaches are not effective in providing co-presence to par-ticipants in a telepresence system due to the lack of general handling of continuous full-body motions including both locomotion and object interaction motions.…”

“…By controlling the avatar, the user can effectively deliver the intention or semantic meaning of their motions to the telepresence participants in a remote space. For example, the local user can make physical contacts with remote participants [OLK * 16, LLL17], point to shared objects [KL20,YYCL21], or interact with environmental elements such as floor or objects [JKK15,KPBL16].…”