I need to integrate some sort of satellite-based positioning for a mobile robot project. I am at the very beginning - I got 2 u-blox GNSS evaluation boards (these NEO-M8Ts) - I chose an integrated antenna and ground plane to get started faster (no external antenna hassle, ground-plane as RTK-ready). I used the u-center software to do some very simple initial tests both measuring stationary and moving positions - I basically only recorded the position from the sensors to a .ubx log file and then played it back.
I am trying to implement this for my mobile robot and I realize that a smartphone-grade GPS (Samsung J5) gives me better preliminary results than an u-blox eval board - I wonder why, I guess Android may fuse the IMU and have better readings even with worse antenna?
What is the correct way of evaluating GPS performance in terms of position accuracy/precision? I know this will depend on the location and its sky clearance (number of sats used, multipath errors etc.) but is there a rule of thumb? I guess I can only evaluate in stationary position since I will have no correct "ground truth" for mobile testing?
Also, I am interested in RTK performance, but that is probably a candidate for another question.