Home, at a glance
Robot location and route around Margaret's home
Hearth robotMargaret— clear route
Battery 78%About 6 hours remaining
Feeling cheerfulSoft green light
Listening gentlySay “Kia ora” or “Help”
⌂
hearthcompanionship, close by
Wi-Fi connected
M
A home companion prototype
Good afternoon, Margaret ☀️
A little company, a little reassurance, and family always close.
A friendly chat
Try “Kia ora”, “How are you?” or “Help”
Hearth · just nowKia ora, Margaret! It's lovely to see you. How are you feeling today?
Things to enjoy together
Small moments for a brighter day
Family photo albumA new photo from Anna
Word of the dayLearn a te reo Māori word
Remember a songShare a favourite memory
Recent moments
A simple shared activity log
Hearth said hello to Margaret
Anna shared a photo from the garden
Robot returned to the living room dock
What we are learning
WEEK 5 · MVP TESTRiskiest assumption: A family member can quickly understand the robot's state and start a reassuring check-in, without feeling the controls are confusing or intrusive.
Clickable test: Invite 3–5 family caregivers to find the robot and start a wellbeing check. Observe their clicks and ask what they expected. Do not explain the interface first.
0 / 0check-in flows completed
Target: 4 of 5 without help, under 30 sec
Target: 4 of 5 without help, under 30 sec
How this prototype works · Architecture & honest limits
Problem-first framing
For older adults living at home, quiet stretches and distance from family can create loneliness and worry. Family members want reassurance and connection without turning care into surveillance. Hearth's value hypothesis: warm, opt-in companionship for older adults and a quick, understandable way for family to check in.
Prototype architecture
- One self-contained browser page (HTML, CSS, JavaScript); no account or server.
- A small graph of home locations routes the robot around rooms; movement is animated in the SVG floor plan.
- Interface events update mock battery, location, Wi-Fi, conversation and activity log state.
- Browser speech recognition and speech output are optional enhancements when available.
Implemented vs simulated
Working interaction: controls, route navigation, emergency flow, call state, family dashboard, check-in counter, activities, bilingual prompt responses and voice-input request.
Simulated: camera feed, two-way audio, video call, robot sensors, person detection, battery, Wi-Fi recovery and navigation. Voice recognition/speech depend on browser support. Māori phrases are a small scripted demo, not a fluent AI model.
Feasibility · desirability · viability
Feasibility: interface flows are buildable; reliable indoor navigation, privacy-safe sensing and accessible speech need a hardware/service PoC. Desirability: validate with older adults and family caregivers using observed behaviour, not hypothetical interest. Viability: still unknown; test who pays, service/support costs and a small caregiver pilot before proposing a revenue model.
Safety & privacy boundary
This is not a medical device or emergency service. It cannot detect falls or contact real people. No camera, microphone, personal data or call is transmitted or recorded. Emergency is a demonstration flow only; a real product needs explicit consent, secure controls, tested escalation and a reliable human response path.
Next experiment
Run a moderated 10-minute prototype test with 3–5 family caregivers and older adults separately. Measure time to locate the robot, successful check-in without coaching, confusion points and whether camera/call controls feel respectful. Revise the riskiest screen, then test again.