Bobo the robo — AR hospital wayfinding

01

The problem

Role: UX research & design
Timeline: 2.5 months
Tools: Figma, Blender, Spark AR

The wayfinding system ends a few metres past the entrance.

Coloured footprints lead away from the front doors and stop before the first junction. Past the lobby every corridor looks the same. Yellow lighting, identical doors, signage written for adults, and nothing that tells a child where they are or which way to turn.

Children almost never navigate on their own. They hold a hand, walk behind a parent, or look at the floor. The one place a child was independent for any length of time was a screen game built into a waiting room wall.

Hospital corridor with coloured footprint stickers leading to emergency registration — Footprints near the entrance

Waiting area and a corridor of identical doors further inside the hospital — Where the trail runs out

02

Observational research

Two afternoons of structured observation.

I ran unobtrusive observation sessions in three zones: the entrance, the Clinic 9 waiting room and the connecting hallways. Notes were timestamped and taken in public areas only. I logged who each child moved with, what held their attention, and any moment they acted independently.

Observation notes by area

Hospital entrance

Children entered accompanied by an adult, looking around the lobby.
Almost every child had a neutral or unhappy expression.

Clinic 9 waiting room

A child was absorbed in a game built into the wall.
He brought his parents over from the seating area to show them how it worked.
The only digital interaction on the first floor, and the only place a child stayed engaged alone for a long stretch.

Hallways

Coloured wayfinding stickers were only near the entrance, not through the building.
Corridors and rooms were dimly lit with yellow lighting and old fixtures.

Across all zones

Most children were accompanied by a female adult.
Many children were in wheelchairs or strollers.
Children stayed physically close to an adult throughout.

Handwritten observation notes describing children in the waiting area — Field notes, Clinic 9

Handwritten observation notes timestamped through the afternoon — Field notes, entrance and exit

What that pointed to

One child took the lead only once in two afternoons, in front of a screen, and the first thing he did was bring his parents over to see it. That is the behaviour the building already produces: a child will lead when something is theirs to hold.

So the guide should belong to the child rather than be read out to them by an adult.

03

Competitive analysis

Built for adults, or built for play. Not both.

I compared Bobo against the two things an average child aged 6 to 12 would otherwise have on a hospital visit: the building’s own static signage, and the apps that exist today. ARWay is a general indoor AR wayfinding platform aimed at adult visitors; Alder Play is a children’s hospital companion app built for comfort and distraction rather than navigation.

The criteria come from what the observation turned up.

Compared against what a child aged 6 to 12 would otherwise use
Criterion	Bobo the robo	Hospital signage	ARWay	Alder Play
A child can use it without an adult reading for them	Intended	No	No	Yes
Routes you to a destination inside the building	Yes	Partly	Yes	No
Guidance continues past the lobby	Yes	No	Yes	n/a
Distance given in a unit a young child can follow	Seconds	No	Metres	n/a
Comfort and distraction content for the rest of the visit	No	No	No	Yes
Runs on a phone the family already has	Yes	n/a	Yes	Yes
Choice of language at first launch	Yes	No	Not known	Not known
Tested with patients in a working hospital	No	Yes	Yes	Yes
Works without new indoor mapping being installed	No	Yes	No	Yes

04

Who it is for

A composite persona: Penelope, age 7.

Penelope is not a real patient. She is a composite built from the observation notes and the age range the project targets, used to keep design decisions anchored to one set of needs: find things without asking for help, and hold some control over a day that has been decided for her.

Portrait of a smiling seven year old girl, standing in for the persona Penelope

Penelope, 7

Lives Toronto
Visit First time in hospital
Device Her mother’s iPhone

Penelope is outgoing and curious, and nervous about a building she has never been in. She is trying to be brave about it. What worries her most is needles and being kept in a bed for a long time, and the part of the day she has any say over is very small.

Goals

Find places without having to ask for help.
Stay calm through the visit.

Needs

Something in the day that feels like hers.
To ask questions and get answers she understands.

Feels now

Anxious
Confused

05

The companion

A character, not an arrow.

The guide had to be something a child would want to follow. Bobo went through several rounds: rounder head, softer limbs, a screen face that can change expression, and finally a colour the child chooses during onboarding so the companion is theirs.

Early mascot sketches beside the first 3D model in mint green — Sketches to first model

Mint prototype next to the final yellow Bobo with a rounded hood and expressive face — First pass to finalRounder silhouette, warmer colour, a face that can react

06

Branding

A system the app and the character both sit in.

The visual language came before the screens: organic blob shapes, a five colour palette, and Sofia Pro at three sizes. The same four colours are the ones offered to the child when they choose Bobo.

Moodboard of playful children's brands, organic shapes and bright palettes — MoodboardReference collected before any mark was drawn

Branding ideaLogo construction, clear space, palette and type scale

07

Wireframes

Onboarding first, then the walk.

The flow splits in two: a short onboarding where the child names themselves and picks Bobo’s colour, and the AR walk where Bobo waves, leads, and announces arrival. Sketching it on paper made the cuts obvious.

The account step came out early. Email and password were crossed off in the second round with the note that it is stressful enough being in the hospital.

Pen wireframes of the onboarding flow: landing, name, patient number, email, password, colour choice — Onboarding

AR walk, profile and search

08

Prototype

Two rounds, and a shorter way in.

The first round asked for a name, a patient number, an email and a password before anything happened. The second round cut the account entirely, added language choice, made the next button stay grey until a real choice was made, and ended with three AR screens that give one instruction at a time.

First prototype: loading screen, four onboarding steps, location category and first AR screen — Round oneFive steps before a child reaches the hallway

Second prototype: language, name, colour, destination and three AR wayfinding screens with turn instructions — Round twoAccount removed, one instruction per AR screen

OnboardingLanguage, name, and Bobo’s colour. Nothing a child would need an adult for.

DestinationRoom, cafeteria, washroom or exit, then a room number if they have one.

The walkBobo stands in the corridor ahead; a single card reads “For 10 seconds, continue straight”.

Arrival“You’re here!” with Bobo celebrating, so the walk ends on something good.

09

Walkthrough

Onboarding through to arrival.

Prototype walkthroughOnboarding through to arrival

10

What I took from it

01

Watch before asking

Nobody told me children never navigate alone. Two afternoons of sitting still in the building did.

02

Every step costs

Cutting the account screens was the single biggest improvement between rounds, and it was a subtraction.

03

Ask, next time

A senior designer said this after I presented the project. Observation told me what children did, not why. The next round should add short consented intercept interviews with parents in public areas.

Hospital wayfinding a child can follow.