1、2023-2-20ChatGPT for Robotics:Design Principles and Model AbilitiesSai Vemprala*,Rogerio Bonatti*,Arthur Bucker,and Ashish KapoorMicrosoft Autonomous Systems and Robotics ResearchThis paper presents an experimental study regarding the use of OpenAIs ChatGPT 1 forrobotics applications.We outline a st

2、rategy that combines design principles for promptengineering and the creation of a high-level function library which allows ChatGPT to adaptto different robotics tasks,simulators,and form factors.We focus our evaluations on theeffectiveness of different prompt engineering techniques and dialog strat

3、egies towards theexecution of various types of robotics tasks.We explore ChatGPTs ability to use free-formdialog,parse XML tags,and to synthesize code,in addition to the use of task-specific promptingfunctions and closed-loop reasoning through dialogues.Our study encompasses a range oftasks within t

4、he robotics domain,from basic logical,geometrical,and mathematical reasoningall the way to complex domains such as aerial navigation,manipulation,and embodied agents.We show that ChatGPT can be effective at solving several of such tasks,while allowing users tointeract with it primarily via natural l

5、anguage instructions.In addition to these studies,weintroduce an open-sourced research tool called PromptCraft,which contains a platform whereresearchers can collaboratively upload and vote on examples of good prompting schemes forrobotics applications,as well as a sample robotics simulator with Cha

6、tGPT integration,makingit easier for users to get started with using ChatGPT for robotics.Videos and blog:aka.ms/ChatGPT-RoboticsPromptCraft,AirSim-ChatGPT code:https:/ rapid advancement in natural language processing(NLP)has led to the development of large languagemodels(LLMs),such as BERT 2,GPT-3