Can you tell us more about how the networking works? Whats the lowest viable bitrate you can remotely operate something like this at? Not like a work-site heavy machine, but say even just your telepresense robot?
On the robot we're using RTP and usually stream the individual cameras separately at about 4Mb each (1440x1080). That is a bitrate where we see little improvement in image quality from going higher. We can go lower, 1-2Mb/stream, but probably if you want to save bandwidth you'll want to process the video into equirect or cubemap or something locally and then stream that at 4K at whatever bandwith your network can handle.
Operating the robot over 4G should work, but I haven't tested it yet :)