Spaces:

Gen-Sim

/

Gen-Sim

like 11

Sleeping

We are a research group. The description of our project is as follows. Hope that we can get the grant to accelerate the rendering speed and improve the demo performance.

We develop an LLM pipeline for generating simulation environments and tasks through program synthesis, as data augmentations for robotic policy learning. The framework consists of three novel components:

1. an automated prompting mechanism that proposes new tasks and implementations for open-world task design
2. a task library for developing more complex simulation environments and data generations
3. a GPT-4 generated incremental benchmark and a language-conditioned multi-task policy training method that leverages the large set of generated tasks, and close the loop on evaluating the task generation pipeline.