Yes, these opinions are based on years of exeperience and graduate research into how to implement control systems using GPS. My masters thesis centered around how ionospeheric effects can degrade the accuracy of GPS based navigation. Granted, my professional experience is not in the world of railroad control, but my personal experience very much gives me insight into how these systems work, since I have learned such from my father, who has more than 40 years of experience in railroad signal engineering. So yes, in my career I have worked on many projects dealing with this exact issue - vital autonomous control.

The biggest concern I have in using GPS in a vital system is the fact it cannot be made failsafe without additional input. I was writing the previous post from my phone, so it was difficult to go into detail, but there are environment effects which are currently unpredictable which will cause GPS to lose accuracy. For example, should there be an increase in solar activity (or even comsic events), GPS can (and does) lose accuracy. This issue is more prevelant around the equator (known as the equatorial anomaly) and at the poles due to the magnetic field of the earth converging (hence the aurora). A simple way to explain this phenomena is imagine a glass of water in with a pencil in it. The light refracts, and the pencil appears to be bent. This refraction (in water) is easy and very simple to model and would not be a problem. But imagine the glass was full of shaken-up sprite. The light reflecting from the pencil will still refract, but having the light pass through a constantly changing medium where the density is not homogenous is nearly imposssible to predict and factor out. This is what GPS signals are passing through. Here in the upper latitudes, this is far less a problem - generally. But what happens if there is a solar event which causes the ionosphere to increase in activity? This causes GPS outages, problems with geolocation (the issue with using it in vital control) and so forth. They are temporary, but what do you do in the meantime?

I have not a problem with using GPS in the system, especially if it is able to provide a sort of "refresh" of the navigational data. None of the systems I have worked with have a single method of gaining nav data (ie the UAV's). They use multiple sources for nav info, such as inertial navigation and so forth. However, I have never seen GPS be the primary nav source. It is always secondary. This means it is not part of the vital control. If it fails, it does not matter. Additionally, GPS was designed by and for the military, with the idea of if it crashes, there is the principle of acceptable losses. To implement such in a commercial setting would be preposterous. Not one loss is acceptable when designing a "failsafe" system.

My concern here is the same as your final statement. Safety is of first importance. For this reason, I have expressed concern over relying upon a system which has inherent capability gaps in it. GPS is cool. It is considered "sexy" by geeks, so if a system is GPS capable, it is for some reason inherently better. My argument is that, while GPS is cool and great, it was not designed to be a primary nav data source for vital control systems and should not be used as such - especial in safety critical applications. Updating these rail control systems is needed. However we will get ourselves into greater trouble if we don't remove the emotional aspect (ie GPS is sexy) and look at it objectively and consider the actual capablities of such a system. There is a phrase I am constantly telling my team of engineers as they go through the design process. Just because an engineer can do something doesn't necesarily mean he should