Lane-sensitive applications

The combination of a digital road map with accurate lane models and an in-car positioning system enables several novel applications [16]. Some require additional parameters and data fields, but all are easily derived from current position traces. These include:

Lane departure warning/lanekeeping

This safety application tracks a car's current offset from the road/lane centerline. If it deviates more than a certain amount, a warning signal could activate or the car could assume control to avoid an accident. The threshold amount could be related to the standard deviation of offsets during typical driving. This application requires very high position accuracy to avoid false positives and negatives.

Lane-level navigation

This enhancement to standard road-level navigation advises the driver as to which specific lane he should choose to reach his destination without excess and last-minute lane changing. Besides the lane models, this application requires a per-lane model of intersection behavior. For example, position traces may indicate that 100% of drivers in the left lane at a particular intersection turn left, 50% of drivers in the right lane go straight, and 50% turn right.

Dynamic lane closures

If aggregate data on lane occupancy is available dynamically though wireless communications, a lane closure application can compare current occupancies for a road segment with past lane occupancies. If that lane is particularly under-represented, the application may infer that it is closed due to an accident or construction. Navigation applications can then take this into account when calculating routes.

We believe these, and other, safety and convenience applications will provide benefits to drivers, and that unsupervised learning from position traces will make them possible without the high cost typically required to build the supporting database by hand.