Nayantra: open-source framework for natural language robot fleet control

Nayantra addresses a key challenge in robotics: the complexity of coordinating multiple robots. Traditionally, fleet management requires deep knowledge of Open-RMF's API and custom dispatcher code. Nayantra introduces a natural language interface using a large language model that communicates via the Model Context Protocol (MCP). This allows operators to issue commands like "Send robot A to delivery point 3 and put robot B on standby" – the LLM parses the intent and automatically calls the appropriate Open-RMF APIs to dispatch tasks. The framework then routes these tasks to Nav2 on real robot hardware or to NVIDIA Isaac Sim for simulation. A critical feature is the stub mode, which lets developers test the entire pipeline locally on a laptop without needing a GPU or a full RMF cluster, making development and testing accessible.

Currently in early open-source release, Nayantra's stub mode works end-to-end and is the easiest way to get started. The integration with real robots and Isaac Sim is functional but still maturing, with the author seeking feedback from the Open-RMF and Nav2 community on task API design and edge cases. This framework has significant implications for robotics developers and operators: it lowers the barrier to entry, allowing teams to manage fleets using natural language instead of writing complex dispatcher code. It also showcases the potential of MCP as a standard protocol for connecting LLMs to existing infrastructure. As the project evolves, it could enable faster prototyping and more intuitive fleet management in warehouses, hospitals, and other multi-robot environments. The open-source nature invites contributions to improve robustness and expand functionality.