Year
2024-25
Role
Lead researcher and first author
Context
IIT Hyderabad
Duration
Sep 2024 to Apr 2025
Team
Supervisor: Prof. Delwyn Remedios (IIT Hyderabad)
Status
Published at ICWIHI 2025 (Springer, AISC series), first author
The problem
Young adults are the most connected generation and often the loneliest, yet mixed-reality assistants are built to help with tasks, not with real situations like staying safe, being understood or meeting people.
The outcome
A systematic review that shows what mixed-reality companions can already do for people in everyday situations, and the four gaps the field still leaves open.
35
papers read in full
ICWIHI 2025 paper
8
research databases searched
ICWIHI 2025 paper
4
open gaps found
ICWIHI 2025 paper
Overview
The goal: before designing a mixed-reality companion, a virtual character that lives in your real surroundings, find out what the research already knows. Here, “navigation” means more than directions. It means getting through real situations: finding your way safely, getting by in a language you don’t speak, and connecting with the people around you, without the companion replacing them.
DISCOVER
Challenge
The need is real. 27% of 19 to 29 year olds worldwide say they feel lonely, and 32% of people in India report a lack of social ties (Gallup, Our World in Data). Mixed-reality assistants already help people work through tasks: guiding a technician, a lesson, a museum visit. Almost none are built for the social and emotional side of everyday life.
Shoulder to shoulder, and everyone alone.
Let’s understand this with a scenario
“Late evening, new city”
Picture a young woman in a new city, late in the evening. She needs to get home safely, ask for help in a language she doesn’t speak, and there is no one she can call. Her phone shows a route. It doesn’t make her feel safe, it doesn’t help her talk to the shopkeeper, and it doesn’t keep her company.
First evening in a new city: a route on the phone, but no one to turn to.
DEFINE
How might a mixed-reality companion help people through real situations, not just tasks, without replacing the people around them?
Today’s systems are either useful or caring, never both. A companion that helps with real life has to be both at once.
Where the research sits: most studies test tasks in short trials; long-term emotional support was not studied in the reviewed work.
DEVELOP
Approach
I ran a systematic review rather than a quick survey of the field, because a survey lets you pick the papers that agree with you and a systematic review makes you account for all of them. Five search terms across eight databases (Google Scholar, IEEE Xplore, SpringerLink, ScienceDirect, Scopus, PubMed, Elsevier, ResearchGate), papers from 2007 to 2024. Studies on VR only, without an assistant in mixed reality, or with no practical evidence were excluded.
Every remaining paper was read in full, checked for bias (sample size, self-reported versus measured data) and sorted into four themes: experience and engagement, isolation and well-being, mental effort in tasks, and technical platforms. I did the scoping, searching, screening, analysis and writing; my supervisor guided the method.
How the review was run: what was searched, what was kept or left out, and how the papers were grouped.
DELIVER
Result
Four findings held across the papers:
characters help, up to a point: people engage more with human-like assistants, but realistic faces feel uncanny, so simple, clean characters work better
body language is underused: gestures and expressions build presence more than talk, yet most systems rely on voice alone
guidance placed in the real space lowers mental effort more than flat on-screen overlays
the evidence that companions reduce loneliness is promising but thin: short-term, small-scale, never long-term
And four gaps remain open: no long-term studies of mixed-reality companions in the reviewed work; almost all research is on headsets, with mobile left untouched; systems are either functional or empathetic, never both; and emotional intelligence is acknowledged in theory but not designed in practice. Those gaps became the brief for my next project, ARiA.
Key takeaway
Being useful is not the same as being a companion. Most systems manage one of the two; people need both.
Reflection
The studies the review covers are mostly short-term and small, so the findings point a direction rather than prove one. Next: long-term studies that follow people for weeks, across phones as well as headsets.
Grounded in
Wang, Smith and Ruiz, Exploring Virtual Agents for Augmented Reality (CHI 2019)
Schmeil and Broll, MARA: a mobile augmented-reality virtual assistant (2007)
Twenge and Campbell, screen time and psychological well-being (2018)
Kenyon, Kinakh and Harrison, social VR, loneliness and social anxiety (Scientific Reports, 2023)
Mittmann et al., augmented reality for social skills: a scoping review (JMIR Serious Games, 2023)
Koleva et al., agent personality and social presence in AR (IEEE VRW 2024)
Full list: in the ICWIHI 2025 paper.
