🧪 New tutorials are being published — build from robot arms to sensors step by step
Skip to content

08 WiFi and the simulation twin ​

This step is optional and not required for self-righting itself.

8.1 Hotspot and status page ​

Hotspot: SSID SR-Support, password srsupport, address 192.168.4.1 (w toggles it; on by default at boot). w is sent over serial: type w in the send box of the serial monitor/assistant and press Enter; serial tools / the Windows UartAssist download are on page 03, section 3.3.

Join the hotspot and open http://192.168.4.1 (phones usually pop the portal automatically). The page refreshes every 300 ms: state, up vector and uprightness bar, joint commands, get-up pose/phase, Home WiFi state and the twin-stream switch. If the browser uses a proxy, add an exception for this address.

8.2 Provisioning (join the home WiFi) ​

On the status page use the Home WiFi card: Scan → pick a network → enter the password → Connect. On success the ESP32 runs AP+STA:

  • the hotspot stays up (bench work is unaffected);
  • the robot is reachable on the home LAN (the page shows its Robot IP);
  • credentials are stored in NVS and rejoined automatically on boot.

2.4 GHz only; scans/connects are refused while a get-up is running.

8.3 Wireless MuJoCo live twin ​

Press Twin ON on the status page (same as serial d), then:

bash
python tools/live_twin.py --tcp <Robot IP>:3333

(In the SR-Support folder; create and activate .venv first as on page 02 - the tool environment is set up there too.)

  • The window mirrors the real attitude live and keeps the body on the floor;
  • The arm is held at its floor-contact angle like the stalled servo (no floor penetration);
  • The USB cable is only needed for flashing and log dumps;
  • Replay a recorded log offline with --replay <file>.

8.4 WiFi telemetry logging ​

bash
python tools/tcp_log.py

Saves the hotspot telemetry stream to logs/wifi_<time>.txt for offline replay and analysis.

8.5 Simulation and training reference ​

The repo ships the MuJoCo model and training code (training/b_line/) for replay and retraining; the model files are described on page 10.


← Back to the tutorial contents