Another facet to this question: Drones specialize in low-altitude, high-resolution remote sensing.
Evaluating these remote sensing technologies requires not just the track of the vehicle, but very solidly constrained GPS points on the ground. Ground Control Points attached to visual targets need to have their position mapped to the best accuracy affordable. Currently, our lab uses an incredibly expensive mapping-grade GPS unit by Trimble that can pretty reliably give 1m accuracy in seconds, but getting every point down to 0.1m or so with postprocessing requires 10 minutes of waiting per point. When there are significant obstructions, like trees, it's even more difficult.
I'm looking into whether it's plausible, instead of using a single expensive unit, to use dozens of very cheap loggers spread out over the area of interest, attached to visual targets. It is basically inevitable that some of them get stolen, so cheap is highly beneficial. The idea would be to leave them all attached to their targets for hours or perhaps even a day, and then average their position later to get high-precision results.
My concerns are price, and whether this would statistically bias the positions in some way. When you get to this level of precision other effects start to pop up... I'm currently struggling with the existential meaning of the WGS84 datum with/without WAAS correction, and whether orbital & physical changes in the Earth would allow generally repeatable measurements.
So what is state of the art, more than three years later, in ultra-cheap, long battery life, basic WAAS-corrected GPS loggers?
The default entry point is presently the $60 Holux M-241.