What it does
ResQ is an AI-powered emergency guidance system for people living alone or disconnected from digital networks. It combines multiple detection methods: a wearable necklace that sends location and alerts when pressed, computer vision that independently detects falls without any user action, conversational AI that assesses consciousness and injury status, and risk-assessment algorithms that determine emergency severity. High-risk situations trigger automatic escalation to caregivers and emergency services, and the system keeps working even without internet connectivity thanks to on-device processing.
How we built it
We developed a browser-based prototype on Replit using HTML, CSS, JavaScript, TensorFlow.js, and Google's MoveNet model. Computer vision runs locally on-device to protect privacy, with MoveNet tracking body posture for fall detection. The system pipes that signal to conversational AI for user check-ins, then combines both inputs into a severity estimate that guides ResQ's response. The physical design includes a solar-powered mobile robot, a Raspberry Pi camera, a wearable necklace, a touchscreen, and a satellite communication module for reliability in areas that lack electricity and internet.
Challenges we ran into
The hardest part was determining context: a camera can detect a person on the floor but cannot tell whether they are resting, exercising, or in a genuine emergency. Traditional alert buttons fail when a user cannot press them. Our solution required integrating several AI systems — computer vision analyzing posture, conversational AI gathering information the camera can't see, and risk assessment combining those signals into a severity estimate. We also had to architect for the digital divide, designing for unreliable internet and power rather than assuming smartphone access.
Accomplishments that we're proud of
We're proud that ResQ goes beyond basic emergency detection to include situation understanding, risk evaluation, and user guidance during a crisis. It addresses my grandmother's real struggle and the similar challenges facing many isolated individuals. Above all, we kept humans involved in critical decisions while prioritizing privacy through local, on-device data processing.
What we learned
Responsible AI extends far beyond accuracy — it means understanding each component's limitations and building protective measures around them. Developing ResQ highlighted the importance of human oversight, privacy preservation, explainable recommendations, and systems that keep functioning despite limited connectivity.
What's next for ResQ
Next, we plan to build the physical prototype — the emergency necklace, solar charging, Raspberry Pi camera, and satellite communication integration. We also aim to enhance the conversational AI, add multilingual capabilities, and conduct real-world testing with users and caregivers to measure how well the system understands emergencies and connects people to support.