Unverified50% confidenceFactExact time
Google's RT-2 demonstrated that vision-language models can directly output robot action instructions.
1
Sources
50%
Confidence
Long-term
Relevance
9/1/2026
First Seen
Sources
Related Entities
Related Claims
Unverified将大语言模型接入机器人的典型范式包括Google的SayCan（用LLM分解任务后调用底层运动原语）和RT-2（端到端生成控制指令）77% similarUnverifiedGoogle DeepMind的RT-2（2023年）将机器人动作词元化为离散token，附加到视觉-语言模型PaLI-X和PaLM-E的输出词汇表中，展现出零样本泛化能力76% similarUnverifiedGoogle于2023年发布RT-2等视觉-语言-动作模型（VLA），将感知、语义理解与操作策略统一在同一Transformer框架中71% similarUnverifiedRT-2, released by Google DeepMind in 2023, was the first to validate the feasibility of fine-tuning robot control on a 55-billion-parameter vision-language model, with generalization capability nearly doubling compared to its predecessor.71% similarUnverifiedAcademia is exploring the application of foundation models in robotics, such as Google DeepMind's RT-2 and Stanford's Open X-Embodiment project.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/839177API
curl https://kongchang.com/api/v1/knowledge/claims/839177MCP
get_claim(id=839177)