Incredible Work on Genesis — Curious About Future Use in Broader Embodied AI Benchmarks

[aroojjaved93]

Hi Genesis team and community 👋,

First of all, hats off to everyone involved in this project. The architecture and modularity of Genesis is incredibly well thought-out. The integration of LLMs, visual grounding, memory, and planning really feels like a leap forward for embodied agents. 🧠✨

I had a couple of open questions and thoughts:

1. Do you plan to expand Genesis to work with other simulation environments** (like Habitat AI, RoboSuite, or Webots) beyond the Minecraft-style settings?
2. How modular is the framework for swapping in alternative vision models or different planning strategies?

I’ve been exploring similar problems on a much smaller scale in automation-focused tasks (e.g., ticket classification and SLA prediction using LLMs + context), and this repo sparked some exciting new ideas.

Looking forward to hearing your thoughts and thanks again for pushing the boundaries in open-source AI research!

Best regards,
Arooj Javed