SIVO.GO

What it does

SIVO.GO replaces multiple single-purpose alerting devices with one portable system. AI recognises emergency sounds, while a separate sensor detects door activity, triggering light, screen and vibration alerts during sleep, showering and travel.

Your inspiration

This began with my lived experience as a profoundly Deaf cochlear implant user. Removing hearing devices to sleep or shower reduces environmental awareness, yet existing products remain fragmented, bulky and expensive. Research with Deaf and hard-of-hearing users showed that many had normalised over a decade carrying several devices or depending on inaccessible hotels. With over 18 million people in the UK affected by deafness, hearing loss or tinnitus, this revealed an unmet need and market pull. I challenged the assumption that every alert requires a separate product and explored how embedded AI could create one travel-ready safety system.

How it works

The prototype uses an Arduino Nano 33 BLE Sense Rev2 with its MEMS microphone, an OLED screen, RGB LEDs, a vibration motor and a real-time clock. I trained a TinyML sound classification model in Edge Impulse using over 500 labelled audio samples, optimised it with the EON Compiler, and deployed it as an Arduino library in nearly 1,000 lines of C++. Model runs locally, recognising fire alarms and doorbells without Wi-Fi or cloud processing. Room, Bed and Bath modes translate alerts into colour-coded light, screen symbols and adjustable haptics. Knocking varied too much for reliable AI classification, so a no-damage vibration sensor provides direct door detection. As urgent sounds are unpredictable, the microphone, processor and model must listen continuously. This increases power consumption, but a safety device cannot depend on intermittent checks; development therefore targets a 1,500–2,500 mAh battery, 8–12 hours of overnight operation, and magnetic charging.

Design process

I mapped lived experience, reviewed products and interviewed Deaf and hard-of-hearing users aged 18 to 70. Priorities: reliable alerts, portability, strong haptics, discretion and accommodation independence. Matrices compared detection, notification, interaction, attachment and charging. Sketches explored home hub, multi-room system and travel kit; users selected compact direction. SolidWorks, Fusion 360 and KeyShot developed form, layout and travel case. Breadboard, perfboard and wired prototypes tested detection, screen, lighting, vibration and modes. 3D-printed housings tested grip, button spacing, one-handed use, under-pillow comfort and packing. Testing exposed two failures. Door-dependent knocks resembled footsteps, tapping or similar sounds, so I replaced AI classification with direct vibration sensing. First interface had excess controls and exposed ports, so I reduced it to tactile mode inputs and magnetic charging. Further iterations proposed stronger haptics, brighter 360° diffuser, anti-slip grip and splash-resistant surfaces. Exploded CAD and DFM refined PC/ABS housings, silicone parts, standard electronics and injection-moulded assembly for pilot manufacture at scale. Bluetooth will sync local time and connect optional door sensor; sound recognition remains offline.

How it is different

Conventional assistive systems divide fire, door and wake-up alerts across separate transmitters, receivers, chargers and fixed installations. Alerting equipment can cost £800-£1,400+ and add 1.5-3 kg; with hearing-device care equipment, travel load may reach 3-5 kg. Phones and smartwatches may be drained, charging elsewhere, too weak to wake a heavy sleeper, permission-restricted or unsuitable for shower use. SIVO.GO challenges both approaches with one user-owned platform targeting under 300g. Its handheld unit can be placed where needed or worn using a lanyard or clip. It requires no Wi-Fi, cloud processing, staff installation or compatible accommodation. Rather than forcing AI onto every signal, it classifies consistent sounds locally while physically sensing variable knocks and door movement. This hybrid approach reduces false alerts, equipment setup and reliance on others, combining personal safety, dignity and portability in one discreet everyday system.

Future plans

SIVO.GO began as a final-year project, but users confirmed demand beyond the study. Despite limited personal funding, I will refine the prototype, develop a PCB, improve reliability and manufacturability, and expand trials with Deaf and hard-of-hearing users, hotels and accommodation providers. I will seek appropriate IP protection, grants, mentors and industry partners for certification, pilot manufacture and market entry. My aim is to apply the latest AI technology and inclusive design to close the gap between conventional medical devices and everyday support, creating an affordable product that improves safety, independence and confidence.

Awards

New Designers: Runner-Up from Stannah Group for Inclusive and Empathetic Design, Cambridge Consultants for Breakthrough Design and Imperial Brands PLC for Healthier Futures.
Degree Show: Winner by JLR for Manufacturable; shortlisted by P&G Gillette for Sustainability, and Cambridge Research Systems Limited for Innovation