assistive device that not just leads people but also offers emotional confidence and security — offering autonomy and confidence via technology.
What it does
Vizhi AI is a wearable AI-neckband powered by AI that employs ESP32-CAM, ToF sensors, and AI models to identify blockages, identify environments, and offer real-time voice guidance via bone-conduction sound. It integrates Generative AI for emotional support and ML-based navigation for turn-by-turn and mapless route guidance.
How we built it
We added ESP32-CAM for vision-based detection, coupled ToF sensors for proximity notification, and employed Python-based AI backend (Gemini API) for contextual notification. The system sends information to a mobile application through Bluetooth, which forwards information between the hardware and the cloud. ElevenLabs TTS is employed for generating natural voice response.
Challenges we encountered
Difficulty integrating multiple sensors due to limited ESP32 memory.
Handling latency between edge and cloud for real-time audio feedback.
Bone conduction headset support and stability of connections.
Optimization of object detection under different lighting and outdoor environments.
Achievements we're proud of
Developed a working MVP combining AI, IoT, and assistive audio technologies.
Designed an intuitive human-centered experience centered around accessibility.
Awarded by industry leaders and chosen for additional incubation and showcase.
What we learned
We studied how to combine hardware and AI for human-centric design, discovered multimodal AI models, and realized how vital empathetic UX is in accessibility technology.
Future directions for Vizhi AI (The Third Eye)
We will:
Augment the model with offline lightweight AI for rural accessibility.
Enhance detection accuracy through LiDAR-based sensing.
Create a cloud platform for caregiver tracking.
Work with NGOs and smart city programs to deploy at scale across the country.
Log in or sign up for Devpost to join the conversation.