What can I use only one ESP32-S3 DevKitC to do?

[dabaosun]

I'm using only one ESP32-S3 DevKiC to evaluate the project. From the http://localhost:3000/ui/index.html, seem the 'Live Demo' and 'Sensing' all not expected. For 'Live Human Pose Detection', it always shows 2 persons, in fact, only 1 person there. For 'Sensing', the classifction always switches in 'PRESENT_MOVING', ‘PRESENT_STILL’, ‘ABSENT’.

So, I'm wondering if the ESP32 mesh is neccessary for all fuctions? and Is this project really deployable and usable in a production environment?

[ruvnet]

What a Single ESP32-S3 Can Do

Great question. With one ESP32-S3 DevKitC, you can:

Works well with 1 node:

• Presence detection — detect if someone is in the room (binary: present/absent)
• Motion classification — PRESENT_MOVING, PRESENT_STILL, ABSENT (this is working correctly for you)
• Basic breathing rate — when a person is stationary and close to the node (~1-3m)
• CSI data collection — raw ADR-018 frames streamed to aggregator

Requires 2+ nodes for accuracy:

• Pose estimation — triangulation needs multiple viewpoints
• Person counting — single-node can't reliably distinguish 1 vs 2 people (which explains the "always shows 2 persons" behavior)
• Localization — needs geometric diversity from multiple positions

Why you see "2 persons" with 1 node:

The multi-person algorithm uses subcarrier group clustering. With a single viewpoint, multipath reflections can create phantom "persons" — the reflected signal looks like a second body. This is a fundamental limitation of single-node CSI, not a bug.

Your setup is working correctly:

The sensing classification cycling between PRESENT_MOVING, PRESENT_STILL, ABSENT is expected and correct for 1 node — it's detecting your presence and movement state.

Recommended 1-node configuration:

# Optimize for single-node presence sensing
python provision.py --port COMX --ssid "YourSSID" --password "pass" \
  --target-ip YOUR_PC_IP --edge-tier 1 --node-id 0

--edge-tier 1 disables multi-person and pose features, focusing on presence + motion which work well with a single node.

Is it production-ready?

• Presence/motion detection with 1 node: Yes, suitable for room occupancy sensing
• Pose estimation: Requires 3+ nodes with calibrated positions — research-grade, not yet production-ready for general deployment
• Person counting: Requires 2+ nodes minimum, 3+ recommended

For a production occupancy sensor, 1 ESP32-S3 with edge-tier 1 is viable today.

[dabaosun]

@ruvnet Thanks for your professional and detailed answer. I will try according to your suggestions.