As someone working across tech and natural disasters, and with a background in engineering, this doesn't make a whole lot of sense from a tech, engineering, or

As someone working across tech and natural disasters, and with a background in engineering, this doesn't make a whole lot of sense from a tech, engineering, or response perspective. High speed mesh networks exist, and commercial operators exist whose tech can cover 160km / 100miles with a single airborne repeater. A repeater on a hill in the distance is enough. Local network without that line of sight, but in bands like Cat-M are good for 16km / 10 miles with a single local repeater. In a mesh configuration 1 mile or 1.6km between meshed nodes is quite achievable. These are all band and power dependent, with differing limitations of each. What is quite clear, however, is that disaster areas are messy, smoky, wet, muddy, windy, inaccessible, chaotic places. Requiring a camera to watch for wiggling requires a clear camera, a high power computer to interpret the dance (at 93% confidence, apparently), and motors at the other end to do the dance. These are all very power hungry things in what is potentially a power poor area. Why would you not look at a mutil band local mesh network with exit nodes and low power requirements already? That's what we are using for human and machine communications, and we have had very few issues with newer tech achieving full data comes in signal deprived areas. When it comes to machine to machine comms it's far more efficient from a data point of view to send the details over the air than it is to do a dance and misinterpret with a dirty camera whilst burning power. This seems like a group of researchers who don't understand the problem space properly and have come up with a pie in the sky solution that is quite impractical. Speaking frankly.