What this demo shows
The robot has three visible interaction states: listening, thinking and speaking. Each state is represented by a different gesture. Rather than using fixed buttons, the HMI exposes the transition logic as draggable nodes that can be moved on the screen to control the state flow.