Collected molecules will appear here. Add from search or explore.
An agentic framework for robotic manipulation that uses LLMs to generate executable code (Code-as-Policies) for controlling robots across simulation (CaP-Gym) and real-world environments.
stars
0
forks
0
The project is an extension or reimplementation of the 'Code as Policies' (CaP) paradigm originally pioneered by Google Research. While it includes a comprehensive gym (187 tasks) and benchmarking tools, it currently has zero traction (0 stars/forks) and is only 8 days old. Frontier labs like Google and OpenAI are the primary drivers of this specific 'LLM-to-code-to-robot' architecture, making this project highly susceptible to obsolescence as official models integrate these capabilities natively.
TECH STACK
INTEGRATION
reference_implementation
READINESS