The AI Team showcases EPFL's top-notch know-hows and technologies through a multidisciplinary semester/year project open to all EPFL students.
Our projects span across diverse fields, including Computer Science, Electronics, and Mechanical Engineering, encouraging a broad spectrum of skills and knowledge.
Objective: Use the current model as a foundation to develop a "Spot" like prototype and implement various models for efficient motion control.
Mechanical & Structural Engineering
Electronics
AI & Machine Learning
Future Vision
Spot® - The Agile Mobile Robot
The goal of this project is to translate sign language (ASL) into natural language.
We work on achieving this through two areas
Computer vision
Using a transformer architecture we detect and classify key-points into sign categories therefore transforming gesture in keywords.
Keywords processing
Using LLMs both locally and through APIs we convert the keywords in natural language.
We plan on iterating on a semester basis, where each member of the GESTURE project team can add something to the pre-existing codebase, whether it is by improving the performances of the pipeline or the number of recognized signs. For the coming semester, a part of the project team will be focused on the deployment of gesture on the robot dog. This means improving the model for it to run on portable devices, and opens the door to multiple new challenges.
AI-generated image
Copyright © 2024 EPFL AI Team - Tous droits réservés.