Scientists developed a system of artificial-intelligence agents that design experimental protocols for laboratory robots and test them before real-world use. A researcher describes their research program and lab setup to the AI agents, which then create step-by-step procedures for the robots to follow. The agents generate a virtual version of the lab where they test the protocols, then direct the physical robots to perform the actual research.
The system addresses limitations in current laboratory robots, which operate with narrow skill sets, limited mobility, and difficulty navigating shared lab spaces.