Large language models (LLMs) like Chat-GPT have been making headlines due to their potential to revolutionize the way we write, learn, and create art by processing vast amounts of text data and using this information to answer prompts. Now, EPFL researchers have applied this technology to robotic design in a case study published in Nature Machine Intelligence.
Josie Hughes, head of the Computational Robot Design & Fabrication Lab at the School of Engineering, Francesco Stella, an EPFL PhD student, and Cosimo Della Santina of TU Delft used Chat-GPT to design a working robotic tomato-harvester. The study provides a framework for humans and LLMs to collaboratively design such devices. The researchers describe opportunities and risks of applying artificial intelligence (AI) tools to robotics, which they argue could change the way we design robots, while simplifying the process.
Hughes said:
“Even though Chat-GPT is a language model and its code generation is text-based, it provided significant insights and intuition for physical design, and showed great potential as a sounding board to stimulate human creativity.”
Potential and pitfalls of AI as ‘inventor’
The researchers and LLM team first discussed the purpose, design parameters, and specifications of their robot in an ‘ideation’ phase. In the second phase, they worked on realizing the robot in the real world by refining LLM-generated code, fabricating the device, and troubleshooting its functioning.
During the first phase, the researchers started with a high-level conversation with the LLM to discuss future challenges to humanity and how robotic crop harvesting could help address the challenge of global food supply. They used the LLM’s access to global data from academic publications, technical manuals, books, and media to provide the “most probable” answer to questions such as “What features should a robot harvester have?”
Once they identified a basic robotic format (a motor-driven gripper for grasping ripe tomatoes), the researchers could then ask more specific questions, such as “What shape should the gripper have?” and request technical suggestions from the LLM, including materials and computer code for controlling the device.
“While computation has been largely used to assist engineers with technical implementation, for the first time, an AI system can ideate new systems, thus automating high-level cognitive tasks. This could involve a shift of human roles to more technical ones.”
But let’s not give any credit to the AI lest it gets cocky
The researchers suggested various other ways of collaborating between humans and AI in their paper, aside from designating Chat-GPT as the ‘inventor’. One of these methods is ‘collaborative exploration’, where AI provides expertise from a wide range of knowledge domains to augment researchers’ expertise. AI can also act as a ‘funnel’, providing technical input and refining the design process, while humans maintain creative control.
However, the researchers warn that each collaboration mode has logical and ethical risks that must be considered carefully. The use of AI raises concerns about bias, plagiarism, and intellectual property, as it’s unclear whether a design generated by AI can be considered original.
“In our study, Chat-GPT identified tomatoes as the crop ‘most worth’ pursuing for a robotic harvester. However, this may be biased towards crops that are more covered in literature, as opposed to those where there is truly a real need. When decisions are made outside the scope of knowledge of the engineer, this can lead to significant ethical, engineering, or factual errors.”
Despite these cautions, Hughes and her team conclude, based on their experience, that LLMs have great potential to be a force for good if well managed: “The robotics community must therefore identify how to leverage these powerful tools to accelerate the advancement of robots in an ethical, sustainable, and socially empowering way.”
Image Credits
In-Article Image Credits
Chat-GTP-designed robot holding a tomato via CREATE Lab/EPFL with usage type - Creative Commons LicenseFeatured Image Credit
Chat-GTP-designed robot holding a tomato via CREATE Lab/EPFL with usage type - Creative Commons License
