Sure, I assume that all cars will have something like that. Heck, since the car will be doing navigation it will likely have found a gas/charging station and pulled over long before it even got to that. But regardless they will never be perfect. What if it sprung a leak and couldn't pull over in time because it judged that there was no suitable shoulder (mountain road, narrow bridge), and this info wasn't in it's database to enable it to plan ahead?
We have been mass producing cars for over 100 years, and by all reasonable measures they have never been as reliable as they are today. Yet they still break down on occasion. Self driving cars will have all the same mechanical and electrical problems that we have today, with software problem on top of that. You can mitigate some of these hardware problems with additional sensors, and fault-tolerant design of the driving computer, but only to the point where the sensors and software are significantly more reliable than the hardware they are monitoring, and only for the situations that are programed for.
There always will be situations where things break down in unexpected ways that the car isn't capable of handling on it's own. And based on the historical rate of reliability improvement, those situations won't be uncommon for quite some time.