Emily Wong, Adélaïde Genay, Jens Emil Sloth Grønbæk, and Eduardo Velloso. 2025. Spatial Heterogeneity in Distributed Mixed Reality Collaboration.
In CHI Conference on Human Factors in Computing Systems (CHI ’25), April
26–May 01, 2025, Yokohama, Japan. ACM, New York, NY, USA, 19 pages.
https://doi.org/10.1145/3706598.3714033
3 Method
The need for a unifying framework arose from our own experience
developing systems that blend heterogeneous spaces in MR. Using a
set of 14 canonical examples [14, 19, 21, 26, 28, 29, 32, 34, 55, 56, 66,
73, 79, 80], we iteratively designed an initial framework proposal
through a series of workshops within the authorship team. Then, we
presented this version at a workshop with 6 additional researchers
specialising in MR. The feedback from this workshop was used to
iterate the framework.
To validate it, we performed a comprehensive literature review
from 1995 to 2024. We used Ens et al.’s [13] systematic literature
review on MR collaboration as an initial dataset covering 1995 to
2019.From this dataset, we identified 80 papers that included systems for remote collaboration. We then supplemented this sample
with papers from the last five years, performing a keyword search
(“MR” OR “mixed reality” OR “AR” OR “augmented reality” OR “telepresence” OR “VR” OR “virtual reality” OR “XR” OR “extended reality”)
AND (“remote” OR “distributed” OR “hybrid”)from 2019 to 2024 in
the same databases as Ens et al. [13] (CHI, CSCW, and ISMAR).
We added any additional papers the authors were aware of and
included a search of IEEE VR proceedings. In total, we identified
345 papers from Ens et al.’s [13] systematic literature review (80), a
search from the last five years of CHI (98), CSCW (6), ISMAR (60)
and IEEE VR (84), and papers supplemented by the authors (17).
Our inclusion criteria were as follows: (1) the system focused
on remote collaboration and (2) it blended two or more dissimilar
physical spaces. We first skimmed the abstract of each record and
excluded any that clearly did not meet the above criteria (302). Then,
we read the full text of the remaining articles (43) and met as a team
to discuss and remove any articles that did not meet the criteria
(11).
This left us with a total of 32 articles, of which 18 were published
in the last five years (2020 to 2024), 8 in the five years before that
(2015 to 2019) and only 6 before 2015. This shows how the topic of
spatial heterogeneity in distributed MR is still relatively small but
has been growing extensively since the 2020 pandemic.
We analysed each system described in the 32 papers according
to our framework. A database of all 32 papers and how they relate
to the framework can be found in our Notion database1 and a hard
copy spreadsheet is also provided in the supplementary materials.
Section 2.3 in the related work gives an overview of key examples
used to demonstrate the framework.
4 The Spatial Heterogeneity Framework
Choosing an MR solution for diverse collaborative tasks is challenging, as activity requirements and spatial heterogeneity often
change during collaboration. Solutions assuming spatial similarity
inherit affordances from the physical environment but struggle
to adapt to highly heterogeneous spaces, limiting their flexibility.
Users may also address spatial challenges by modifying the layout
of their physical space or adapting tasks rather than switching MR
solutions.
To address the lack of a theoretical basis for designing and comparing systems that overcome spatial heterogeneity, we propose
the Spatial Heterogeneity Framework (see Figure 1, right). This
framework externalises abstract concepts, enabling researchers
to better discuss, compare, and develop systems for collaboration
across dissimilar spaces.
The framework comprises four components: activity zones, the
heterogeneity ladder, blended proxemics, and an MR solutions matrix.
and debate the challenges of blending heterogeneous spaces. A
practitioner can start with any of the outer components (Figure
1, right), depending on their constraints and priorities, and then
move inward to consider the heterogeneity ladder before moving
out again to consider the remaining outer rectangles. The worksheet in the appendix provides a more detailed step-by-step process
for this approach. At a high level, the components are defined as
follows:
(1) Activity zones(Section 5), which are used to define the
roles of each area of the physical space in the collaborative activity.
This component identifies the “blended zones” in each space to be
mapped to each other.
(2) The heterogeneity ladder(Section 6) describes the level of
physical similarity between the remote spaces’ respective blended
zones. The level of similarity between blended zones of distributed
spaces depends on the layout of the rooms’ fixed (e.g. walls), semifixed (e.g. a desk) and mobile (e.g. whiteboard pens) features [46].
The ladder serves as a hierarchical metaphor, where each step
up the ladder represents an increase in spatial heterogeneity—the
degree of dissimilarity between distributed spaces. As the level of
heterogeneity increases, it becomes more challenging to create a
sense of shared space across distributed locations.
(3) Blended proxemics(Section 7), which refers to the social
interactions between people, objects and space supported across the
distributed locations. For example, in Figure 1 (left), an MR blending technique might adjust Alice’s avatar in Bob’s space so that her
pointing gesture aligns correctly with the table in Bob’s environment, ensuring the proxemic cues are meaningful and contextually
appropriate.
(4) MR solutions matrix(Section 8) enable blended proxemics
that are not already afforded by the direct mapping between the
distributed spaces. For example, a system might use re-directed
