ChatGPT-Based Robot Design Could Revolutionize Future of Agriculture

Researchers from the Dutch TU Delft and EPFL in Switzerland have harnessed the power of advanced natural language processing AI platforms to design robots. Leveraging the capabilities of ChatGPT, they successfully developed a tomato harvesting robot, forging a valuable collaboration between engineers and AI.

According to Assistant Professor Cosimo Della Santina and PhD student Francesco Stella from TU Delft, along with Josie Hughes from EPFL, ChatGPT provided invaluable insights during both the conceptual and realization phases of the design. The chatbot suggested automating the economically most valuable crop and offered practical advice, such as using a silicone or rubber gripper to protect the tomatoes from damage. Additionally, ChatGPT recommended a Dynamixel motor as the most suitable option to power the robot.

ChatGPT acted as a researcher

The researchers found the joint design process to be positive and enriching, although they acknowledged that their role as engineers evolved. They took on more technical responsibilities while ChatGPT acted as a researcher and engineer. This human-AI collaboration has yielded an efficient tomato harvesting robot.

While highlighting the benefits of working with ChatGPT, the research team also recognizes the associated risks of relying solely on AI for robot design. The output of ChatGPT can potentially be misleading and generate incorrect information or biases, as AI models are trained to provide the most probable answers without human validation. The team also raises important considerations such as plagiarism, traceability, and intellectual property when working with AI models.

AI is going to accelerate robot development

Despite the challenges and risks, the researchers are enthusiastic about the future possibilities of utilizing AI, like ChatGPT, in robot design. They continue to employ the tomato harvesting robot for further research purposes and are exploring the autonomy of AI in designing its own body. The ultimate goal is to leverage Large Language Models (LLMs) such as ChatGPT to support robot developers in addressing the 21st-century challenges while preserving creativity and innovation in the field of research.