A robot scientist that discovers quantum materials

Qumus is a robotic laboratory system that combines large language models with physical manipulation to conduct autonomous materials science research. The system performs the complete scientific workflow—hypothesis generation, experimental design, execution, and analysis—without human intervention. It has demonstrated AI-driven creation of graphene and fabrication of atomically thin field-effect transistors, marking the first autonomous realization of these achievements. The system uses closed-loop feedback and autonomous error correction to refine experiments in real time, establishing a framework for embodied AI systems that learn directly from physical experimentation.