This project consists of developing an end to end toy with a customizable look and advanced conversational, empathetic and interactive features. Equipped with low cost sensors, displays and microcontrollers, Sahar will house lightweight machine learning algorithms to interact in Arabic and in a personalized way with normally developed and neurologically different kids.
Challenges
The educational and entertainment market don’t offer yet any cost effective smart, personalized and interactive toys. Because early years’ learning and play are essential for the development of the future generation, our Create-X project aims to develop a unique toy for kids to interactively play with. Sahar differentiates itself from existing products in few aspects and will face the following challenges. It is intended to support Arabic speech and text and should appeal to kids with Arabic roots. Its appearance should be customizable with an option for folklore outfits. Sahar’s sensors and on board AI modules should enable it to communicate offline with kids and acquire their reactions in a real time manner for better personalized interaction and behavioral therapeutic analysis if needed. Sahar should be recording all play sessions and yet maintaining learner’s privacy as data won’t be shared without the explicit consent of the child caregiver. Sahar is designed to be the smart toy that every kid will want to have and every kid could afford and that is the biggest challenge.
Methods
Sahar aims to democratize smart toys and bring equity for underprivileged and mentally and socially challenged kids. From a hardware perspective
Sahar will be equipped with low cost sensors such as cameras
proximity sensors
speaker and microphone to sense and interact with kids. Sahar will be designed to operate either in an offline or in an online manner
while preserving the privacy of the user’s data. Microprocessors and storage will be needed to run AI algorithms optimized for low resource environments. From a software perspective
Sahar will rely on SOTA deep learning algorithms that will be particularly adapted for Arabic speech and text understanding and generation. Given the unpredictable environments of kids’ play and developing articulation
Sahar will rely on common sense AI
multimodal sensed data in conjunction with historical data for its interaction with the kids. Sahar will leverage federated learning to enable lifelong learning for the AI software powering it. In this manner
Sahar will be fine-tuned in a personalized manner to the style and need of the kid interacting with it. It will also have an online option to inform caregivers about the play session highlights and report any unusual activities.
Academic Majors of Interest
Computer Science
Electrical and Computer Engineering
Computer and Communication Engineering
Education
Psychology
Business
Graphic Design
Preferred Skills
Coding
Hardware Design
AI
Human Machine Interaction
NLP
Education
Occupational Therapy
Neuroscience
Graphic Design
Early Child Education
Behavioral Analysis
Business
Software Development
Students
Join today
Apply to the project today, and join other students and faculty members.
