How can LLMs transform the robotic design process?

被引：23

作者：

Stella, Francesco ^{[1
,2
]}

Della Santina, Cosimo ^{[2
,3
]}

Hughes, Josie ^{[1
]}

机构：

[1] Ecole Polytech Fed Lausanne, CREATE Lab, Lausanne, Switzerland

[2] Delft Univ Technol, Dept Cognit Robot, Delft, Netherlands

[3] German Aerosp Ctr DLR, Inst Robot & Mechatron, Wessling, Germany

来源：

NATURE MACHINE INTELLIGENCE | 2023年 / 5卷 / 06期

关键词：

Compendex;

D O I：

10.1038/s42256-023-00669-7

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

We show that large language models (LLMs), such as ChatGPT, can guide the robot design process, on both the conceptual and technical level, and we propose new human–AI co-design strategies and their societal implications.

引用

页码：561 / 564

页数：4

共 15 条

[1] Anonymous, 2023, NAT MACH INTELL, V5, P1, DOI [10.1038/s42256-023-00613-9, DOI 10.1038/S42256-023-00613-9]
[2] Brants T., 2007, P 2007 JOINT C EMPIR, P858
[3] Budzianowski P., 2019, PREPRINT
[4] Chen Mark., Evaluating Large Language Models Trained on Code, P2021
[5] GPT-3: Its Nature, Scope, Limits, and Consequences
Floridi, Luciano
Chiriatti, Massimo
[J]. MINDS AND MACHINES, 2020, 30 (04) : 681 - 694
[6] Can AI invent? Comment
George, Alexandra
Walsh, Toby
[J]. NATURE MACHINE INTELLIGENCE, 2022, 4 (12) : 1057 - 1060
[7] Huang W., 2022, PREPRINT
[8] Ichter B., 2023, PMLR, P287
[9] Ramesh Aditya, 2022, arXiv preprint arXiv:2204.06125, V1, P3, DOI DOI 10.48550/ARXIV.2204.06125
[10] Shanahan M., 2022, PREPRINT, DOI DOI 10.48550/ARXIV.2212.03551

← 1 2 →