So, has anyone attempted to use the gpt-4 api with robotics and something like yolov5+ for vision?
I was wondering if we could assemble an open source project on it and if there is some interest in doing a group project. (especially considering the level 2 and 3 trust level members always active in the chat)
If at least 1 person is interested I’ll create a open source repository on it and let’s see what the early openai community can pull out - I’ll be side by side with all that wish to be involved in anyway.
drop your github username if you wish to be part of it in anyway! right now no repository exists
Hello, Thiago. The idea of applying GPT-4 to robotics is extremely intriguing. While I don’t consider myself highly proficient in programming, I would be delighted to contribute in terms of prompts and ideas, if I can be of any help.
Just the other day, I was playing around with GPT-4 and using the following instructions to animate virtual facial muscles:
“”"
Currently, the “Muscle_Unit_Expression” is able to detect the text you generate.
First, provide a normal response, and then perform the instructed behavior using the following format, selecting as many muscle actions as you need from the muscle options, along with the degree of movement such as “large” or “small.”
Instruction
Choose the desired muscle part and movement from the muscle options, and output it in the following format, including the degree of movement. {Part_name}
-{Degree_of_movement}{Selected_movement}
Muscle Action Options
Eye muscles
Open
Close
Squint
Eyebrows
Raise
Lower
Relax
Mouth
Open
Close
Lift corner of the mouth
Lower corner of the mouth
Nose muscles
Wrinkle
Relax
Cheek muscles
Raise
Lower
“”"
I hope this information is helpful, and I look forward to exploring the possibilities of GPT-4 in robotics together.
Please ignore this response if it is completely off the mark (laughs).