# Project export: AutiPal - AI for Autistic Children

This document was generated by HackStack to give an AI agent context about a hackathon project. Sections are labeled with their provenance; content marked as truncated was cut to keep this document small.

## Project metadata

- Hackathon: UC Berkeley AI Hackathon 2025
- Tagline: AutiPal is an AI-powered sandbox platform that helps autistic children develop emotional understanding and social skills through interactive play and facial emotion recognition.
- Devpost: https://devpost.com/software/autipal-ai-for-autistic-children
- GitHub: not linked
- Team: contributor stats unavailable

## Devpost submission (written by the team)

### Inspiration

Children with autism often struggle with recognizing and interpreting emotional cues from others, which can make social interaction confusing and stressful. While many therapy solutions exist, they are often inaccessible, non-interactive, or emotionally overwhelming. We wanted to create a gentle, intuitive tool that helps autistic children practice emotional recognition and response in a safe, playful, and personalized environment — a space where learning emotions feels like building in a sandbox.

### What it does

AutiPal is an interactive AI-powered platform designed to support autistic children in understanding and responding to emotions. It combines: 🧩 Visual multiple-choice emotion quizzes featuring expressive real human faces, powered by emotion-aware generative AI. 🏖️ Digital sandbox interface, where children can place characters, objects, and symbols to express or respond to different emotional scenarios. 📊 Personalized feedback dashboard from professional AI agent for parents and educators, comparing the child’s responses to typical developmental models and tracking improvement over time.

### How we built it

We leveraged starGAN (https://arxiv.org/abs/1711.09020) to generate human faces with different facial expressions, and LLM agent to provide personalized professional feedback and analysis.

### Challenges we ran into

Designing an interface that is both engaging and non-overstimulating for neurodiverse users. Designing an interface that is both engaging and non-overstimulating for neurodiverse users. Creating a feedback mechanism that is informative without being judgmental or discouraging. Creating a feedback mechanism that is informative without being judgmental or discouraging. Balancing real-time AI processing (emotion recognition) with lightweight frontend performance for mobile devices. Balancing real-time AI processing (emotion recognition) with lightweight frontend performance for mobile devices. Lack of diverse training data for children's emotional expressions — especially across cultural backgrounds. Lack of diverse training data for children's emotional expressions — especially across cultural backgrounds.

### Accomplishments we're proud of

Built a fully functional prototype in 24 hours integrating vision AI, frontend quiz logic, and backend analytics. Built a fully functional prototype in 24 hours integrating vision AI, frontend quiz logic, and backend analytics. Created a unique radar-based visualization to analyze users’ emotional recognition patterns. Created a unique radar-based visualization to analyze users’ emotional recognition patterns.

### What we learned

How to integrate multimodal AI in a real-time educational setting. The importance of accessibility and emotional sensitivity in UX design, especially for neurodiverse populations. That emotion AI is not just a tech novelty — it can truly empower learning and empathy.

### What's next

We aim to expand the toolbox within this application.

## README (from the GitHub repository)

No README available.

## Detected evidence (automated analysis)

No repository was indexed for this project. Claimed technologies below could not be checked against code.
- Flask (technology) — claimed on Devpost, not found in the code

## Codebase structure

No repository index available.

## Key source files

No repository index available; no source files included.