UC Berkeley Researchers Introduce Learnable Latent Codes as Bridges (LCB): A Novel AI Approach that Combines the Abstract Reasoning Capabilities of Large Language Models with Low-Level Action Policies

A novel AI approach called LCB combines the abstract reasoning capabilities of large language models with low-level action policies in robotics. LCB overcomes the limitations of existing methods and demonstrates proficiency in diverse reasoning and long-horizon tasks.