This was a polish week. Next week I finally point a camera at the robot and film it, so everything I did was about making NEURA feel finished instead of almost finished. No big new systems, just closing the gaps that would look rough on video.
The headline is hands-free voice control. I can change the eye color just by talking. If I say go emerald, or even can you do pink, NEURA snaps the eyes to the closest color it knows and answers with a quick one-liner. It makes the change feel like a choice NEURA made, not a setting I toggled.
Video works the same way. Tell it to play a video and it fills the round face, say close video and the eyes come back. On a round screen that transition looks better than it has any right to.
The trickiest call was movement. I gated every drive command behind the actual word drive, so normal talk can never make the body roll away by accident. I learned that one the hard way, and a robot wandering off mid-sentence is not the vibe.
Expressions got tripled to 37, and every emotion now moves both the eye shape and the color together. Angry glows red, happy goes gold, and it reads instantly. I also gave NEURA the ability to look at its own screen and describe what is on it, using Claude vision, which is eerie in the best way.
The rest was cleanup: killed the white screen that flashed on reboot, made the Bluetooth headset reconnect on its own, and hid the mouse cursor for good. Small things, but they are the difference between a prototype and a product. NEURA is genuinely demo ready. Time to film.