Unverified50% confidenceFactExact time
Google 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
1
Sources
50%
Confidence
Long-term
Relevance
7/14/2026
First Seen
Sources
Vizuara机器人学习课程笔记获取指南与免费资源推荐
redditr/deeplearning7/12/2026
Related Claims
UnverifiedRT-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.79% similarUnverified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）70% similarUnverifiedAcademia is exploring the application of foundation models in robotics, such as Google DeepMind's RT-2 and Stanford's Open X-Embodiment project.69% similarUnverifiedGoogle DeepMind's RT-2 and robotics projects such as Figure AI have invested enormous costs in collecting human behavioral data.68% similarUnverifiedGoogle DeepMind提出了ReAct（Reasoning + Acting）代理循环模式67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/503254API
curl https://kongchang.com/api/v1/knowledge/claims/503254MCP
get_claim(id=503254)