基座模型技术背景下的具身智能体综述
|
李颂元,朱祥维,李玺
|
Survey of embodied agent in context of foundation model
|
Songyuan LI,Xiangwei ZHU,Xi LI
|
|
表 6 机器人操控的基座智能体比较 |
Tab.6 Comparison of foundation agent for robotic manipulation |
|
智能体 | 语言 | 基座模型 | 真实世界 | 多具身 | 参数量 | SayCan[131] | ✓ | PaLM | ✓ | ✓ | 5.40×1011 | R3M[132] | ✓ | — | ✓ | — | — | Gato[54] | ✓ | — | ✓ | ✓ | 1.2×109 | RT-1[53] | ✓ | SayCan | ✓ | — | 5.40×1011 | PaLM-E[11] | ✓ | PaLM, ViT-22B | ✓ | ✓ | 5.62×1011 | RT-2[55] | ✓ | PaLI-X | ✓ | — | 5.5×1010 | SMART[60] | — | — | — | — | 1.08×107 | MVP[58] | — | — | — | — | 2.2×107 | Real MVP[59] | — | — | ✓ | ✓ | 3.07×108 | RPT[63] | — | — | ✓ | ✓ | 3.07×108 | RoboCat[57] Yang 等[133] | — — | — — | ✓ ✓ | ✓ ✓ | 1.2×109 1.80×108 |
|
|
|