Only one can be detected

[13331112522]

i have flashed the latest 0.5.0 to three esp32 nodes and use provision to distinguish them by --node-id * like:

python firmware/esp32-csi-node/provision.py --port /dev/tty.SLAB_USBtoUART
--ssid "wifi" --password "psw" --target-ip 0.0.0.0
--node-id 0,1,2

But the server can only detect one node anyway.

[colonelblacc]

--target-ip 0.0.0.0 — this is likely wrong,each ESP32 needs to know the IP of your server to send UDP packets to. Use your machine's actual local IP (e.g. 192.168.1.x). You can find it with ipconfig on windows (Windows) and check if a port conflict occurs.

[ruvnet]

Two issues here:

1. --target-ip 0.0.0.0 is wrong (as @colonelblacc noted)

Each ESP32 needs to know the IP address of your server/computer to send UDP data to. 0.0.0.0 means "send to nowhere." Find your computer's actual LAN IP:

# macOS
ifconfig | grep "inet " | grep -v 127.0.0.1

# Example output: inet 192.168.1.20

Then provision each node separately:

python firmware/esp32-csi-node/provision.py --port /dev/tty.SLAB_USBtoUART \
  --ssid "wifi" --password "psw" --target-ip 192.168.1.20 --node-id 0

python firmware/esp32-csi-node/provision.py --port /dev/tty.SLAB_USBtoUART \
  --ssid "wifi" --password "psw" --target-ip 192.168.1.20 --node-id 1

python firmware/esp32-csi-node/provision.py --port /dev/tty.SLAB_USBtoUART \
  --ssid "wifi" --password "psw" --target-ip 192.168.1.20 --node-id 2

2. --node-id 0,1,2 is not valid syntax

Each node must be provisioned individually (one --node-id per flash). You flash node 0, unplug it, plug in node 1, flash node 1, etc.

For a 3-node TDM mesh (better accuracy), also add --tdm-slot and --tdm-total:

# Node 0 of 3
--node-id 0 --tdm-slot 0 --tdm-total 3

# Node 1 of 3
--node-id 1 --tdm-slot 1 --tdm-total 3

# Node 2 of 3
--node-id 2 --tdm-slot 2 --tdm-total 3

Make sure you're using the latest firmware: v0.5.0-esp32

[13331112522]

I have flash 3 nodes as following with node id=0,1,2:
python firmware/esp32-csi-node/provision.py --port /dev/tty.SLAB_USBtoUART
--ssid "wifi" --password "psw" --target-ip 192.168.10.139
--node-id 2 --tdm-slot 2 --tdm-total 3
and use the latest firmware:0.5.0

BUT Only 1 ESP32 Node Detected with Node id=1 Despite Having Multiple ESP32s
My Setup:
2-3 ESP32-S3 devices running wifi-densepose firmware
Each configured with different node IDs (0, 1, 2) using --node-id flag
Docker container running wifi-densepose sensing server
ESP32s transmitting WiFi CSI data via UDP to Mac (192.168.10.139:5005)
Root Cause Identified
The --node-id command only changes runtime configuration, NOT persistent storage.

When ESP32s reboot, they revert to the default node_id: 1 stored in flash memory.

[ruvnet]

This is a confirmed bug that was fixed in v0.5.0-esp32.

Root cause

Prior to v0.5.0, --node-id was stored only in runtime memory, not persisted to NVS (non-volatile storage). When ESP32s rebooted, they all reverted to node_id: 1, so the server saw 3 devices reporting as the same node.

The fix (v0.5.0+)

PR #279 changed provisioning to write node_id to NVS flash. Now each ESP32 retains its unique node ID across reboots.

To resolve

1. Flash all 3 nodes with v0.5.0-esp32
2. Re-provision each node — the new firmware writes node_id to NVS:

   # Node 0
   python provision.py --port /dev/tty.SLAB_USBtoUART --ssid "wifi" --password "psw" \
     --target-ip 192.168.10.139 --node-id 0 --tdm-slot 0 --tdm-total 3
   
   # Node 1
   python provision.py --port /dev/tty.SLAB_USBtoUART --ssid "wifi" --password "psw" \
     --target-ip 192.168.10.139 --node-id 1 --tdm-slot 1 --tdm-total 3
   
   # Node 2
   python provision.py --port /dev/tty.SLAB_USBtoUART --ssid "wifi" --password "psw" \
     --target-ip 192.168.10.139 --node-id 2 --tdm-slot 2 --tdm-total 3

3. Verify by rebooting each node and checking serial output — you should see node_id: 0, node_id: 1, node_id: 2 respectively after reboot.

This should resolve the "only one detected" issue. Each node will now maintain its unique identity persistently.

[amoscndc-tech]

这是确认存在的bug，已在v0.5.0-esp32中修复。

根本原因

在 v0.5.0 之前，仅存储在运行时内存中，未持续存在于 NVS（非挥发性存储）。当ESP32重启时，它们都恢复到，所以服务器看到3台设备报告为同一个节点。--node-id``node_id: 1

修复方案（v0.5.0+）

PR #279更改了配置，改写为写入NVS闪存。现在每个ESP32在重启后都保留其唯一的节点ID。node_id

解决

1. 把所有3个节点刷入v0.5.0-esp32

2. 重新配置每个节点——新固件会写入 NVS node_id：# Node 0
   python provision.py --port /dev/tty.SLAB_USBtoUART --ssid "wifi" --password "psw"
   --target-ip 192.168.10.139 --node-id 0 --tdm-slot 0 --tdm-total 3

   Node 1

   python provision.py --port /dev/tty.SLAB_USBtoUART --ssid "wifi" --password "psw"
   --target-ip 192.168.10.139 --node-id 1 --tdm-slot 1 --tdm-total 3

   Node 2

   python provision.py --port /dev/tty.SLAB_USBtoUART --ssid "wifi" --password "psw"
   --target-ip 192.168.10.139 --node-id 2 --tdm-slot 2 --tdm-total 3

3. 通过重启每个节点并检查串行输出来验证——重启后你应该会分别看到 ，这些数据。node_id: 0``node_id: 1``node_id: 2

这应该能解决“只检测到一个”的问题。每个节点现在都会持续保持其唯一身份。

Could you please explain how to train a model? Is it that without training the model, we cannot effectively track and detect human skeletons?

[13331112522]

Actually i flashed with latest 0.5.0 version at the first place, and I will try again later on and report the results. Thanks