Skip to content

Latest commit

 

History

115 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 

Repository files navigation

🍎 VisionAssist — HackHarvard 2025

🏆 Built for HackHarvard 2025

VisionAssist is a next-generation Apple Vision Pro application that redefines spatial interaction through gesture recognition, real-time environment mapping, and AI-driven context awareness.
Built on VisionOS, it bridges the gap between human intuition and spatial computing — merging augmented reality, machine learning, and computer vision into a single immersive platform.


⚙️ Technical Overview

VisionAssist leverages the Apple Vision Pro’s advanced sensor array and VisionOS SDK to interpret user gestures, facial cues, and real-world objects. The system integrates spatial mapping, hand tracking, and scene understanding to enable natural, touchless interaction in 3D space.

Core innovations include:

  • Spatial Anchoring: Persistent digital overlays attached to real-world coordinates.
  • Hand Gesture Recognition: Dynamic gesture input powered by VisionOS frameworks and ML models.
  • AI Context Engine: On-device neural model that predicts user intent and optimizes UI flow.
  • Realtime Raycasting: Detects object boundaries for precise interaction with virtual interfaces.

🧠 Features

  • Gesture-Based Navigation — Control apps and content through natural motion and hand signals.
  • 🌍 Spatial Awareness — Uses LiDAR and VisionOS spatial mapping to render AR overlays in real-time.
  • 🧩 Adaptive UI Layer — Interfaces dynamically adjust based on gaze, gesture, and context.
  • 🔊 Seamless Multimodal Input — Integrates gaze, gesture, and voice for a unified interaction model.
  • On-Device Inference — AI runs locally for low-latency feedback and privacy preservation.

🧰 Tech Stack

Component Technology
Platform VisionOS (Apple Vision Pro)
Language Swift / SwiftUI
Frameworks RealityKit, ARKit, CoreML, Vision
AI/ML On-device CoreML models (gesture classification & object recognition)
3D Rendering Reality Composer Pro / SceneKit
Backend (optional) Python / FastAPI microservice (for analytics or model updates)

🧩 System Architecture

+--------------------------+ | VisionAssist (VisionOS) | +-----------+--------------+ | v +-----------+--------------+ | Vision + CoreML Engine | | - Gesture Recognition | | - Scene Understanding | +-----------+--------------+ | v +-----------+--------------+ | RealityKit UI Renderer | | - 3D Overlays | | - Spatial Anchors | +-----------+--------------+ | v +-----------+--------------+ | Backend API (optional) | | - Data sync / analytics | +--------------------------+


🧪 Installation & Setup

Requirements

  • macOS 14.0+ with Xcode 15+
  • Apple Vision Pro device or VisionOS simulator
  • VisionOS SDK installed via Xcode

Running the App

# Clone the repository
git clone https://github.com/yourusername/VisionAssist.git
cd VisionAssist

# Open the Xcode project
open VisionAssist.xcodeproj

# Build & deploy to Vision Pro / VisionOS Simulator

📈 Impact

Designed and prototyped under HackHarvard 2025, demonstrating real-world applications of spatial AI and mixed reality.

Showcases how Apple Vision Pro can revolutionize education, accessibility, and productivity through gesture-first interaction.

Developed with a focus on low-latency UX, energy efficiency, and on-device ML inference.

🧑‍💻 Team & Credits

Developer: Gary Samuel, Eason Ying, Kyle Kelly, Nancy Pazzi
Event: HackHarvard 2025 — Harvard University
Category: Spatial Computing / AI Interaction
Location: Cambridge, MA


🎓 HackHarvard Prestige

This project was conceptualized and built during HackHarvard 2025, one of the world’s most prestigious university hackathons.
It embodies the event’s mission — innovation through interdisciplinary technology, merging AI, AR, and human-centered design to define the future of spatial computing.


---

Would you like me to make a **README variant** focused more on **“pitch-style storytelling”** (like what you’d show HackHarvard judges during demo day — elegant, slightly marketing-driven, and visually structured)?  
That version reads like a **startup launch page**, not just a technical doc.

About

Lock in

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages