Term Project Product Ideas — AST01.43 User-Centric Technology Design
Research notes for our group, 10 Oct 2026. ภาษาไทย
What the professor wants
- A real IT product that is already sold (not a research process), analysed with the frameworks from class.
- It must be genuinely new, something the professor probably hasn't seen.
- No "mask swaps": the same idea with a new provider or casing (e.g. another bike-sharing service) will be rejected.
- Course outline criteria: Innovative · Concise · Applies what we learned · User-centric design (20% of the grade, presentation only).
How we checked each product
- Not a mask swap. No existing product does the same job the same way.
- IT is the core. AI, software, data or a connected service does the main job.
- Probably new to the professor. Our proxy: launched or commercialised recently, CES honoree, not in mainstream Thai media. This is a guess.
- Documented development story. Founder origin, prototypes, tests, so we can fill the 5 Design Thinking stages with real evidence.
- A clear human persona.
The kind of product that fits. Halter, a solar GPS collar from New Zealand that moves cows with sound and vibration ("virtual fences"). Its public history fills every Design Thinking stage: the founder grew up on a dairy farm, the first prototype was a belt and a lunchbox, and a test cow named Twinkle showed they needed speakers on both sides. On top of the persona and Design Thinking, we should also show the Hook Model and the 5 Elements of UX, which our professor tests.
Top picks
1. .lumen Glasses for the Blind — accessibility (Romania)
- What it is: a headset that guides blind people while walking, as a replacement for a guide dog.
- How it works: depth cameras and on-device machine learning build a 3D map of the street in real time, like a self-driving car ("pedestrian autonomous driving"). A haptic pull gently steers the user's head toward a safe path. It works offline, needs no pre-mapped area and doesn't talk constantly.
- Who uses / who pays: blind pedestrians use it. List price is about €9,999, but the first ~1,500 orders were paid through Romania's government assistive-tech programme, so the user is not the customer.
- Why it's new: Be My Eyes still needs a human volunteer on a video call. .lumen replaces the guide dog with AI. CES 2026 Innovation Awards Honoree.
- Story for Design Thinking: founder Cornel Amariei grew up in a family where everyone except him had a disability, and he led innovation at Continental (automotive). Founding insight: about 40 million blind people but only ~20,000 trained guide dogs.
- Sources: AI for Good (ITU) · Interesting Engineering · The Gadgeteer · JA Worldwide: Cornel Amariei · TYN Magazine, CES 2026
2. Dryad Silvanet — environment and disaster (Germany)
- What it is: an early-warning system for wildfires.
- How it works: small solar-powered "electronic nose" sensors are hung on trees. They sniff gases such as hydrogen and carbon monoxide from a fire that is still smouldering, often within the first 30–60 minutes, before smoke is big enough for cameras to see. Each sensor runs a little AI and passes messages along a LoRa mesh network to a gateway.
- Who uses / who pays: forest agencies, park authorities and landowners. More than 50 deployments, including Thailand, Lebanon, Italy, Canada and the USA.
- Why it's new: most systems watch for smoke with cameras (e.g. Pano AI). Dryad smells the fire earlier, a different mechanism.
- Bonus: links to the LoRa topics in our AST01.47 wireless course.
- Sources: tech.eu · IoT Evolution · Environment+Energy Leader
3. SwingVision — sports (USA)
- What it is: an app that turns one iPhone or iPad into a tennis and pickleball referee and coach.
- How it works: AI tracks the ball in 3D from a single phone camera. It gives line calls, keeps score, counts shot stats and cuts highlight clips automatically. No installed stadium cameras, unlike Hawk-Eye.
- Who uses / who pays: amateur players and coaches, US$179.99/year for Pro.
- Why it's new: professional-level officiating and stats from a phone instead of a multi-camera system.
- Best Hook Model of all our candidates: stats and highlight clips after every match are the variable reward, and match history plus the improvement trend is the investment.
- Story: founder Swupnil Sahai came from Tesla's Autopilot team.
- Sources: Founder interview · Pulse 2.0 (funding) · App Store
4. Sanas — work and call centres (USA)
- What it is: real-time AI accent translation for call-centre agents.
- How it works: while an agent in India or the Philippines speaks, generative AI re-voices the speech into a US or UK accent almost instantly, keeping their tone and emotion. It also cancels background noise and can do live language translation.
- Who uses / who pays: agents use it; outsourcing companies (BPOs) pay. Used by 12 of the world's top 20 BPOs (2024).
- Why it's new: speech-to-speech accent conversion, not subtitles or ordinary noise filters.
- Discussion angle: a built-in ethics debate. Does it erase identity, or protect agents from accent-based abuse?
- Sources: Sanas · CX Today · Intellyx
5. Dusty Robotics FieldPrinter — construction (USA)
- What it is: a robot that prints the building plan directly onto the construction floor.
- How it works: it reads the digital building model (BIM) and drives across the concrete slab printing full-size wall lines, doors and markings. It replaces workers with chalk lines and tape measures: about 10× faster, accurate to 1/16 inch. More than 100 million sq ft printed.
- Who uses / who pays: construction contractors and layout crews.
- Story for Design Thinking: founder Tessa Lau (machine-learning PhD, IBM Research, then service robots) visited construction sites to find where AI could help, a ready-made Empathize stage.
- Sources: AEC+Tech · Tessa Lau (Wikipedia) · For Construction Pros
6. XRAI AR One — accessibility and ageing (UK)
- What it is: glasses that show live subtitles of real conversations for people who are deaf or hard of hearing.
- How it works: the phone app listens, speech AI transcribes, and the text appears on the lenses. It tells speakers apart, supports 140+ languages and saves a searchable transcript. US$1,449 (launched June 2024).
- Who uses / who pays: people with hearing loss, often older people, and their families.
- Story: founder Dan Scarfe built it after his 97-year-old grandfather struggled to follow family conversations. This fits the ageing-society theme.
- Watch out: Hearview makes similar glasses (about US$2,000), so present XRAI as the original and Hearview as the competitor.
- Sources: Speech Technology Magazine · Fox News (Hearview)
Backups
UMITRON CELL — aquaculture (Singapore/Japan)
- What it is: a smart feeder for fish farms.
- How it works: an underwater camera watches the fish eat. An AI "Fish Appetite Index" scores how hungry they are and automatically adjusts feeding. Farmers can watch and control it from a phone. It is solar powered and holds 400 kg of feed.
- Story: founder Ken Fujiwara was a satellite engineer at JAXA who noticed that Earth is mostly ocean.
- Results: 10–20% feed cost saving, and 20% lower feed-related emissions (company-reported).
- Risk: very close to Halter (AI for farm animals), a known past example, so the professor may see it as the same idea.
- Sources: Japan Gov "Kizuna" · UMITRON · Aquahoy interview
Elemind — sleep (USA, MIT)
- What it is: a headband that helps you fall asleep without drugs.
- How it works: EEG sensors read your brainwaves, and the headband plays bone-conduction sound timed to those waves in a real-time loop to nudge the brain toward sleep. In a small study, people with insomnia fell asleep 10–15 minutes faster. About US$349 plus an optional app membership.
- Story: from MIT Media Lab research (Ed Boyden); US$12M seed funding.
- Watch out: Dreem (France, ~2017) did something similar, so check novelty.
- Sources: MIT News · IEEE Spectrum · Tom's Guide
Other products we checked
| Product | What it does | Why it's not a top pick | Source |
|---|---|---|---|
| Tevel (Israel) | Tethered flying robots with vision AI that pick only ripe fruit; sold as a service | Farming again, close to Halter; still in grower trials | AgFunder |
| Beewise BeeHome (Israel) | Robotic hive with cameras and AI that feeds, medicates and heats bee colonies | Animal farming, too close to Halter | Farm Progress |
| Pano AI (USA) | 360° mountaintop cameras plus AI spot wildfire smoke; ~US$50k/yr | Smoke cameras already exist; good as Dryad's comparison | MIT Tech Review |
| Nobi (Belgium) | Ceiling lamp with AI vision that detects falls, asks "are you OK?", and calls family | Fall detection already exists (e.g. Vayyar radar) | TechHive |
| Kohler Dekoda (USA) | Camera on the toilet rim; AI reads stool for gut health | Throne and Withings U-Scan do similar things | Wareable · Fitt Insider |
| Winnow Vision (UK) | Camera above kitchen bins; AI recognises wasted food and its cost | Leanpath does similar | The Spoon |
| Greyparrot (UK) | Cameras over recycling conveyors identify every item by material and brand | Business-only, harder persona | Recycling Product News |
| Augury (Israel) | Sensors "listen" to factory machines; AI predicts breakdowns | Crowded market, company from 2011 (great origin story though) | Qualcomm Ventures |
| Lucid Bots Sherpa (USA) | Drone that washes high-rise windows | Mostly hardware | TechCrunch |
| Osmo (USA) | AI predicts how molecules smell; "teleported" the scent of a plum | Sold to fragrance brands; hard persona | NBC |
| Hypershell (China) | Wearable exoskeleton with AI that adds leg power while walking | Several similar exoskeletons; mostly hardware | Gizmodo |
| Nuance Audio (Italy/France) | Hearing aid hidden in glasses frames | A hearing aid in a new body | HearingTracker |
| Bird Buddy (Slovenia/USA) | Bird feeder with a camera; AI names the bird species | From 2022, now widely copied (great Hook example though) | TechCrunch |
The first five links we started from
| Product | What it does | Verdict | Source |
|---|---|---|---|
| Hyodol | Korean talking robot doll with an AI chatbot for seniors living alone; medication reminders and inactivity alerts | Possible, but close to ElliQ (US companion robot) in a doll body | Gigazine |
| Be My Eyes on Meta glasses | Blind users video-call a volunteer through Meta's smart-glasses camera | Weak: a famous app moved onto glasses = mask swap | Be My Eyes |
| Tractive | GPS and health tracker collar for dogs | Weak: crowded pet-tracker category | Neowin |
| Eight Sleep Pod | Water-cooled mattress cover that auto-adjusts temperature from sleep data | Possible, but well known in tech circles | Sleep Foundation |
| Breville eyeQ | Toaster with a colour sensor that stops at your shade | Weak: a sensor, not an IT product | GadgetGuy |
How to choose
- Want a strong Hook Model? (It was on the midsem.) Choose SwingVision.
- Want "the professor has surely never heard of this"? Choose Dryad Silvanet or Dusty Robotics.
- Want the strongest all-rounder? Choose .lumen.
- Want to keep the ageing theme? Choose XRAI AR One.
- Before committing, ask the professor after class: "Is this new enough?"
Some figures are company-reported or from older articles; check before putting them on a slide.