pragmatic

• commute schedule prediction
  Gather empirical data about time-in-motion, so that I have solid data about the probability-curve of how long my morning commute takes, with different curves for different starting times, different expected weather, etc. (most people do this in their heads, but the system will be able to more quickly accomodate new paths or incorporate new conditional events)
• commute route optimization
  Gather empirical statistical data about how long alternative routes take, allowing me to more quickly and accurately optimize routes than can be done with a human alone. ("alternative routes" can include taking significantly different lanes at key bottlenecks on very large expressways)

knowledge for its own sake

• driving time-waster quantifier
  Gather empirical data on how long I spend stopped at stop lights, and how much time is lost due to slower-than-maximum traffic.
  (At one level, these things aren't a "waste of time" because they're an inherent required part of the current system. However, "inherently necessary to this process" does not equate to "inherently necessary to arriving at the goal via any method". Thus, this data goes a small way towards quantifying how much one might potentially save by switching to a completely different transportation method)