DeepYardDeepYard
E

ELITE

Self-improving embodied AI framework that learns from mistakes and transfers knowledge across tasks

Open SourceFree

About

ELITE (Experiential Learning and Intent-aware Transfer) is a research framework that enables embodied AI agents to learn from physical interactions and past failures. Unlike traditional vision-language models trained on static datasets, ELITE allows agents to improve through real-world experience and transfer learned knowledge across different robotic tasks. Designed for researchers working on embodied AI and robotics applications where continuous learning from environmental feedback is critical.

Details

Language
Patterns

Tags

autonomousframeworkopen-sourcemulti-agentpythonragmemory