Remote connection for robots - VPNs over Cellular - MAX BR1 5G

Hey folks,

I’m looking for general advice to start, and then specific advice on how to implement it.

We deploy robots to various industries outdoors, so we depend on cellular + starlink, usually using the MAX BR1 5G.
I pay for Peplink PrimeCare, SpeedFusion, etc, happy to pay to get this use case “solved” for me.

My required use case is this:

  • The modem is on a cell network, and provides DHCP local addresses to devices on the LAN (ex: A lidar)
  • I need to be able to talk with this sensor, as if I am on the same LAN (so a VPN or similar), even though I am on a totally different network in my home or workplace.

Here’s the options I see:

  1. Peplink’s InTouch
    • Works for simple cases, breaks down in strange ways.**
  2. Peplink’s SF Connect
    • This just seems like onboard multiwan agregation, does not allow ingress
  3. Peplink’s SpeedFusion VPN
    • Peplink to peplink connection, might be robust and simple to use, but requires a peplink on both ends… not ideal.***
  4. OpenVPN Client on Router to a server in cloud
    • This seems to not work well, despite trying a lot. It never seems to connect
  5. OpenVPN server on router, client is my laptop
    • This doesn’t work, since on cellular the modem IP address I get isn’t reachable (expected), and DynDNS doesn’t see it either.
  6. Run Tailscale (or similar) in a container on the Peplink
    • Works, if I have a very modern Peplink, but doesn’t work for older models.

** InTouch breaks down for cases that use file uploads, websockets to a significant degree, re-directs, etc. Lots of our devices do this.
*** This is feasible… but means my engineering team all need Peplinks at home, or have a jump-box in the office, which is cumbersome, and suffers from high complexity.

Am I missing something? Is there some other way to acheive my required use case?

I tried hosting my own OpenVPN server, and though I can get my laptop to see and work with it just fine, the Peplink provides no logs I can see to troubleshoot why it fails to ever connect. Does anyone have a simple example config that would fit my use case?

Going further, the ideal case is that any of my engineers can run some simple script to connect to a robot and troubleshoot it, as if on the same LAN. So whatever soltution there is, making it easy to re-configure from Peplink to Peplink is a very nice to have feature.

I know I could run Tailscale on a computer in the same robot network, and use that as an exit node, but one of my use cases is being able to do disaster recovery when the main robot computer is down, and only the Peplink is up and working.

To re-iterate, I am happy to pay $$ to solve this problem.
But if the only solutions is option 3. above, doesn’t that mean that I’ll need 1 Peplink / engineer? So with a team of 50 people and 100 robots (for example), I need 150 Peplinks? Good for product sales I guess… :stuck_out_tongue:

Thanks for your advice!

I solved something very similar for an Ag customer recently. Their issue was intermittent and unreliable cellular connection so not sure that’s the same problem you’re trying to solve. In my particular case I replaced the cellular connectivity portion with Rajant Kinetic Mesh (think your BR1-5G is expensive - lol). So 10 robots each with a teltonika router used the WAN port to attach the mesh node. The mesh nodes all terminated at a peplink router (BR1-5G) with uplink over Starlink and the patchy Cellular. I didn’t know too much about the robots, but I can tell you all this. The robots were located in Far North Queensland. The Robotics engineers were remotely accessing each platform from Pennsylvania USA. Each robot spewed its lidar, telemetry and video camera feeds that the Ag folk and US based Engineers demanded. In this case the robots were using Teltonika routers so as far as peplink goes you’re ahead of the game.
A different project was going south with the peplink and starlink in Aus not being accessed reliably by some folks in California. I recommended they grab a peplink device and run a SF instance with a L2 tunnel between Aus and USA. Now their teleremote truck appears as if it was plugged into a port of the Cali router.

Moving along I think a way ahead would be to host a Fusionhub instance in the cloud… You can have multiple. End points running off the Fusion Hub instance… As long as your engineers can access the hub, they’ll access the routers. If that doesn’t work a physical peplink can indeed become your hub… My personal preference is for the physical. But that’s me…

1 Like

Really appreciate your insight!