Toborino is a small wheeled robot built with Arduino, sensors, and electronic components. It can move, detect obstacles, perform emergency stops, and communicate with a web dashboard in real time.

In this talk, we will explore adding a new natural language control mode: commands like “move forward slowly”, “explore the room”, or “stop if you encounter an obstacle” are transformed by an LLM into structured, validated actions and then dispatched to the robot.

The goal is not to hand over unchecked control to AI, but to use natural language as an interface while preserving safety, clear rules, and predictable behavior.