Can ChatGPT design bots?

Researchers from Delft University of Technology and the Swiss Technical University, with the help of chatGPT, developed a tomato picking robot after careful consideration and acceptance of the challenge.

# In a test environment, a tomato picking robot designed by researchers from TU Delft and EPFL was moved using ChatGPT. Image credit: Adrien Buttier / EPFL

ChatGPT specializes in poetry, essays, books, and potential bot designs. However, there are risks and benefits to using ChatGPT during the design process.

Researchers from TUDelft and the Swiss Technical University EPFL have developed a tomato harvesting robot with the help of chatGPT. After discussions with ChatGPT, they decided to take up the challenge and design a bot.

Helpful Advice

The researchers adopted the design decisions of ChatGPT, especially focusing on its input during the concept phase. The researchers emphasize that ChatGPT expands designers' knowledge in various areas, such as identifying economically valuable automated crops. During the implementation, ChatGPT provided some helpful advice, such as “Use silicone or rubber as grippers to prevent tomatoes from being crushed” and “Use Dynamixel motors for optimal robot drive.” The result of the collaboration is a tomato-harvesting robotic arm that demonstrates the synergy of humans and artificial intelligence (AI).

ChatGPT as a researcher

The researcher believes that the collaborative design process is active and enriching. However, the engineer's role shifted to technical tasks. In unusually extreme cases, AI can take full control of the robot's design process, providing all necessary input, while humans follow its lead without question. In this setup, Large Language Models (LLMs) are the researchers and engineers responsible for all technical aspects. At the same time, humans take on the role of managers, responsible for defining design goals and overseeing the process.

Risk of Misinformation

The current LLM cannot implement such extreme scenarios, and its desirability is questionable. Unvalidated LLM output can be misleading, creating the risk of misinformation and bot bias. Working with an LL.M. raises concerns about plagiarism, traceability and intellectual property. The team plans to use tomato harvesting robots in their ongoing robotics research. They also explore the role of LLM in designing new robots, focusing on AI autonomy in shaping the body. The researchers conclude by highlighting the open question of using LLMs to assist robot developers while retaining the creativity and innovation necessary to meet the challenges of 21st century robotics.

The above is the detailed content of Can ChatGPT design bots?. For more information, please follow other related articles on the PHP Chinese website!