NIMO Controller lets humans and AI agents command self-driving labs

Self-driving laboratories (SDLs) promise to accelerate scientific discovery by automating experiments, but developing the orchestration software remains technically demanding. Existing frameworks are primarily designed for human interaction and lack standardized interfaces suitable for AI agents. To address this, researchers Naruki Yoshikawa and Ryo Tamura propose a novel SDL software architecture based on the Model Context Protocol (MCP), where all SDL functionalities are exposed through MCP servers. Following this principle, they introduce NIMO Controller, an MCP-based SDL orchestrator that provides a visual programming interface automatically generated through MCP-based tool discovery. This allows human users to design experimental workflows without writing any code, while the same MCP backend can be directly accessed by AI agents, creating a unified interface for both humans and autonomous systems.

The team validated NIMO Controller through a case study on a color-matching SDL, confirming the usability of the MCP-based architecture. The system successfully demonstrated how a single orchestration layer can serve both human scientists—via a drag-and-drop visual workflow—and AI agents that query the same capabilities programmatically. This work significantly lowers the technical barrier to building and operating self-driving labs, making them accessible to a broader range of researchers. By standardizing SDL component interaction through MCP, NIMO Controller paves the way for more interoperable and collaborative autonomous laboratories, where human intuition and AI efficiency can work side by side.