Skip to main content

Network Configuration

Choose the network layout before running on hardware. The important decision is where the sim2real processes run and which network carries the Pico stream.

Wired Laptop Deployment​

Use this layout when the laptop runs all code. The laptop talks to G1 over Ethernet and talks to Pico over the lab Wi-Fi.

If you choose a ZMQ bridge mode from Robot I/O, run the bridge on the laptop:

uv run scripts/real_bridge.py

Use ip -br link to find the interface name, and add --interface <laptop_ethernet_interface> only when the Ethernet interface is not the default eth0. Pico should stay on the lab Wi-Fi so the laptop and Pico can communicate through the lab network.

External Wi-Fi Deployment​

Use this layout when all runtime code runs on the onboard Orin, and both the Orin and laptop join the lab Wi-Fi. The laptop is only the operator console over SSH. Pico also joins the lab Wi-Fi and communicates with the Orin through that network.

SSH from the laptop into the Orin, then run the deployment commands on the Orin. Use this mode when the lab Wi-Fi is stable enough for Pico traffic and SSH control.

Orin Wi-Fi Deployment​

Use this layout when all runtime code runs on the onboard Orin, and the Orin provides its own hotspot through an external Wi-Fi adapter. The laptop and Pico both connect directly to that hotspot.

Create the hotspot on the Orin with the setup script:

warning

Any operation that changes G1 network interfaces should be done with an Ethernet cable connected to G1. Keep the cable path available until the new Wi-Fi path has been verified.

bash scripts/setup/setup_g1_hotspot.sh \
  --interface wlan1 \
  --upstream wlan0 \
  --ssid hdmi-deploy \
  --password hdmi1234

The defaults create the hotspot on wlan1 with 10.42.7.1/24 and route client traffic through wlan0. Connect the laptop and Pico to the hotspot. Pico and Orin then communicate directly through the Orin Wi-Fi adapter instead of the lab Wi-Fi.

Manual Built-In G1 Hotspot With Laptop Internet Egress​

Use this layout when the external-adapter hotspot layout is not practical. The previous layout needs an extra Wi-Fi adapter on G1, but the G1 Type-C socket is easy to damage and may not reliably hold a dock.

The intended behavior is manual: G1 boots in normal Wi-Fi client mode, you SSH in through the normal Wi-Fi path, then you run one hotspot command on G1. That command switches the built-in G1 Wi-Fi interface into AP mode. The laptop then joins that hotspot and you reconnect to G1 through the g1-hotspot SSH host.

Configure the laptop SSH alias once:

Host g1-hotspot
  HostName 10.42.7.1
  User elijah

Boot G1 normally and SSH into it through the existing Wi-Fi client path, for example:

ssh g1-rp
cd ~/sim2real

Then run the hotspot command on G1. Pass the built-in G1 Wi-Fi interface name, for example wlan0:

warning

This step switches the G1 built-in Wi-Fi interface from client mode to hotspot mode. The current Wi-Fi SSH session may disconnect. Keep an Ethernet recovery path available while changing this setup.

bash scripts/setup/setup_g1_hotspot_via_laptop.sh \
  --interface wlan0 \
  --ssid hdmi-deploy \
  --password hdmi1234

This profile is intentionally not configured to autoconnect. After a reboot, rerun the script or manually start the profile with sudo nmcli con up hdmi-g1-ap-via-laptop only when the laptop gateway is available.

On the laptop, connect the laptop's USB Wi-Fi adapter to the G1 hotspot and enable NAT through the laptop's upstream internet interface:

bash scripts/setup/setup_laptop_g1_gateway.sh \
  --wifi-interface <laptop_usb_wifi_interface> \
  --upstream-interface <laptop_internet_interface> \
  --ssid hdmi-deploy \
  --password hdmi1234

The defaults use 10.42.7.1/24 on G1 and 10.42.7.2/24 on the laptop USB Wi-Fi adapter. The G1 default route points at 10.42.7.2. The laptop keeps its own default route on <laptop_internet_interface> and only NATs the G1 hotspot subnet through that interface.

After the laptop joins the hotspot, connect to G1 through the hotspot address:

ssh g1-hotspot

After setup, verify both the local link and internet egress:

# On the laptop:
ping 10.42.7.1
ip route get 8.8.8.8

# On G1:
ping 10.42.7.2
ping 8.8.8.8

If G1 can reach 10.42.7.2 but cannot reach 8.8.8.8, check that the laptop --upstream-interface is the interface that actually has internet access, and that laptop firewall rules allow forwarding.

To stop using the built-in Wi-Fi hotspot and return the interface to normal Wi-Fi client use, bring the hotspot profile down:

sudo nmcli con down hdmi-g1-ap-via-laptop
sudo systemctl stop hdmi-g1-ap-via-laptop-dnsmasq.service