Cave

A Quest for Co-Located Mixed Reality: Aligning and Assessing SLAM Tracking for Same-Space Multi-User Experiences

“…The most basic means of alignment is to perform a one-point alignment, a one-shot transformation from the device's current posi-tion+orientation in reality to a specified position+orientation in the virtual coordinate space which has a congruent, equivalent position+orientation in reality. This approach has typically been seen in projects such as CAVE and CAVRN [20,23], where each audience seat in reality has an equivalent position noted in virtuality, such that the MR device can be aligned with said point, giving the audience member the equivalent view in MR as in reality, and such a transformation can be generated trivially: This alignment approach is simple but effective -you need only define a point in virtuality ('target'), and an equivalent point in reality that your device can track; e.g., aligning a headset or controller with this point, or tracking a fiducial marker/QR code/spatial anchor at this point, so that we have a position and orientation in the virtual coordinate space that we can align with the known position/orientation of the 'target' in reality. Multiple devices can be aligned in this way, meaning that, regardless of differences in their individual coordinate spaces, the eventual position of the devices will be aligned, allowing for co-located MR.…”

“…The most basic co-located application of this approach is in the creation of seated audience experiences -alignment with reality is achieved merely by ensuring that audience seat locations in MR mirror their physical arrangement and positions precisely in reality. In exemplar scene (1 Audiences.unity), we provide a basic mock-up of audience seating for an immersive co-located VR experience such as CAVE [23]. We have created 6 GameObjects denoting 6 alignment points, one for each seat (see Figure 3), each with attached OnePointAlignment components.…”

“…Equally, if the user performs the calibration by aligning an MR device directly, and the device is in the wrong orientation, they would find that the MR world would be offset by the difference between the expected and actual orientation at alignment. CAVE [23] appears to have used an operator-driven approach, and it seems likely that the headsets were placed in their pre-determined seat positions for calibration prior to audience arrival, or that audience members were asked at the start to look straight ahead prior to calibration. If the headset has an available front-mounted camera feed (not possible on the Oculus Quest, for example), alternatively placing a fiducial marker/QR code on the space in front of the realworld seat would allow for headsets to auto calibrate themselves, however any inaccuracy in detecting the orientation/position of the marker would again impact alignment.…”

“…For play, research into co-located augmented playspaces (CAPs) [34] has established benefits in terms of stimulating physically active behaviour, social interactions, cognitive development in children, and fundamentally in enabling unique, joyful experiences. And in art, co-located MR performances such as CAVRN [20], Holojam [15], and CAVE [23] have pushed the boundaries in terms of audience experience, with "the adaptation of both cinematic and theatrical elements offer[ing] a unique set of affordances for content designers and producers" [23]. Such experiences might incorporate multiple types of MR devices and displays (the utility of which was particularly emphasised by ShareVR [17]) and multiple user roles, from passers-by, to spectators, to headset users [37].…”

“…However, these solutions are often "black boxes" -closed source solutions, introducing tight coupling to specific MR SDKs, often relying on internet-dependent cloudbased APIs that introduce concerns regarding costs and privacy, with interoperability across platforms often limited by design as a means of creating a 'walled garden' for a given device ecosystem. Conversely, whilst co-located experiences have been repeatedly examined in research without relying on such dependencies [9,16,20,23], the methods by which aligned experiences (from seated audiences to arena-scale co-location) were facilitated have never been sufficiently documented to enable replication.…”