Research Project
GLOW: Generative AI and Embodied Learning
for Omni-Intelligence World Model
Research Team
Lead Researchers:
- Dr. Enno Huang
- Dr. Adam Kaplan
- Dr. Nhut Ho
Collaborators:
- Michel Ingham
- Tiago Stegun Vaquero
- Juan Delfa
- Steven Myint
Student Team:
- Damir Kozhamkulov
Funding
- Funding Organization:
- Funding Program:
SYNOPSIS
The GLOW project builds an Omni-Intelligence World Model with Large Foundations.
- Embodied learning through real and simulated visuomotor experience, and sim-2-real transferring
- World model that predicts future states to enable robotics manipulation and navigation anticipation in Lunar and Mars missions
- Task planning for in-context learning via large robotics foundation model (e.g., π0.7)
- Embodied AGI will make the multi-robotics coordination in autonomous AGI modeling
Research Questions & Research Objectives
Research Methods
- Modeling: Glow autonomous system and general embodied AI world models that integrate robotics action, language proficiency, visual perception, trustworthy memory, intuitive reasoning and Gen-Adaptability.
- Tasks: This project aims to build a Large Embodied Foundation Model for autonomous rover navigation, robotics manipulation, multi-robot collaboration, and execution.
- Training Process: Glow experiments with π0.5-DROID on the CADRE rover serve as baseline measurements and input signals.
Research Deliverables and Products
Commercialization Opportunities
- Scale multi-robot swarms coordination.
- Incorporate multi-modal chain-of-thought reasoning.
- Establish the imitation learning feedback loop with Gen-AI.
- Generate the foundation model dataset and the related benchmark in Moon and Mars.
Research Timeline
Start Date:
End Date: TBD
Research Team
Lead Researchers:
- Dr. Enno Huang
- Dr. Adam Kaplan
- Dr. Nhut Ho
Collaborators:
- Michel Ingham
- Tiago Stegun Vaquero
- Juan Delfa
- Steven Myint
Student Team:
- Damir Kozhamkulov
Funding
- Funding Organization:
- Funding Program: