Unverified50% confidenceFactExact time
Google DeepMind的RT-2（2023年）将连续的末端执行器位姿和夹爪开合量量化为256个离散区间，附加到视觉-语言模型PaLI-X和PaLM-E的输出词汇表中
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
Unverified2022年DeepMind发布了TALM，Meta发布了Toolformer，探索让模型自主决定何时调用外部API，但依赖微调67% similarVerifiedGoogle DeepMind's RT-2 was the first to directly use a vision-language model to generate robot action commands.67% similarUnverifiedDeepMind launched the Auto-RT project, combining large language models with the robot control model RT-2, using the LLM as a task planning layer to decompose natural language instructions into low-level action sequences.66% similarUnverified早在2022年，DeepMind的RLAM和Meta的ToolFormer论文就已论证了语言模型自主调用外部工具的可行性65% similarUnverifiedGoogle DeepMind's SIMA project trained a general-purpose agent capable of executing natural language instructions across multiple 3D games.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/506760API
curl https://kongchang.com/api/v1/knowledge/claims/506760MCP
get_claim(id=506760)