A small robot with a local brain

Reachy Mini Jetson Assistant is a project for Reachy Mini Lite that combines a microphone, camera, spoken responses and expressive movement. NVIDIA’s July project roundup describes it as running locally on Jetson Orin Nano Super, without cloud APIs or an internet connection at runtime.

The footage shows the robot beside its camera view and control interface. This is a developer build with separate hardware and setup, rather than a ready-to-use feature in every Reachy Mini.

The clever part is fitting everything

The Jetson AI Lab tutorial describes an 8GB setup that balances several jobs: understanding speech, interpreting camera frames, generating a response and speaking while moving. Its choices include a smaller quantized vision model and a lightweight serving setup.

That makes the project a useful example of a broader design problem: a robot needs its whole conversation to fit, not just a model that can load once. A system that handles a still image may behave differently when speech and movement run alongside it.

What to look for in the next demo

Our suggested checks are simple: can it recover when someone interrupts, keep track of the right person, and finish a response without an awkward pause? Those observations would tell you more about living with a desk assistant than a single polished answer. The linked tutorial provides the build details; this coverage does not claim an independent performance test.

Explore the original source ↗

Source published 2026-07-28. Coverage is based on the maker’s announcement and demonstration.