Device Connection
Adding a device
Q: How do I add a Pico-G1?
- In the DEVICE TREE area on the left
- Click the
+button in the upper-right corner - Fill in:
- Device name:
Pico-G1or a custom name - IP address: default
192.168.49.10
- Device name:
- Click Save
Q: The board IP is not 192.168.49.10
If the board has changed networks or uses a custom IP:
- Find the actual IP via the router admin page / serial console
- Edit the device → update the IP → save
Q: The device stays red (not connected) after being added
Troubleshoot in this order:
- Network: from the PC,
ping 192.168.49.10; if that works, continue - Board status: check power and the Ethernet link LED
- On-board SSH service: the board must have SSH enabled (port 22)
- On-board user: Pico-G1 firmware ships with a
shimetaduser (no password)
Connection status
Q: Green dot vs red dot?
| Color | Meaning |
|---|---|
| 🟢 Green | Connected, board reachable |
| 🔴 Red | Not connected |
Q: The device suddenly went from green to red
Possible causes:
- The board rebooted
- Network fluctuation
- On-board SSH service failure
What to do:
- Click the Reconnect button on the device card
- Or restart the IDE to let it reconnect automatically
Q: The resource monitors on the right all show 0?
The board is not reachable. Go back to "The device stays red after being added" and troubleshoot.
On-board users
Q: I forgot the default root password
You need to re-flash the official board firmware (which resets the root password).
Re-flashing wipes all data on the board. Back up your important models under
/mnt/first.
SSH configuration
Q: Is the SSH port 22?
Yes. Pico-G1 official firmware uses SSH port 22 by default.
Q: Can I change the SSH port?
Technically yes (edit /etc/ssh/sshd_config), but the IDE connects to port 22 by default; after changing it the IDE can no longer connect automatically.
Q: The board IP changed
Edit the device → update the IP → save.
Multiple boards
Q: Can I manage several Pico-G1 boards at once?
Yes. Add each board separately; IPs must not conflict.
Q: Can multiple boards run different models in parallel?
Each board runs inference independently. But a single board can run only one shimeta-svp or shimeta-xmm at a time.
Networking
Q: The PC and the Pico-G1 are on different subnets — does it work?
Yes, as long as:
- The PC can ping the board
- Not being in the same broadcast domain does not affect the IDE (the IDE does not rely on broadcast)
Q: Can I connect to a board outside my LAN remotely?
Yes (public IP or NAT traversal), but:
- Keep latency ≤ 100 ms (higher makes SSH unstable)
- Board credentials travel in plaintext; a VPN is recommended
On-board services
Q: What services run on the board?
ps | grep shimetaKey services you will see:
| Service | Purpose |
|---|---|
| shimetad_ | The user process the IDE connects to over SSH |
| shimeta-svp | 9 scene detection services (started at runtime) |
| shimeta-xmm | Custom model inference service (started at runtime) |
| shimeta-py | Python runtime (used when editing code in the IDE) |
Q: An on-board service crashed
# Restart SSH
/etc/init.d/sshd restartQuick troubleshooting
| Symptom | Solution |
|---|---|
| Auto-discovery can't find the board | Check cable, IP, subnet; temporarily disable the firewall |
| Test connection times out | Board not powered / loose cable |
| Can't connect with the on-board user | Wrong firmware version; flash the official firmware |
| Board suddenly disconnects | Use the reconnect button; check power stability |