Physically Consistent Map Matching.

An important data source for traffic analysis is GPS trajectory data. However, due to measurement inaccuracies, such data does not necessarily align well with data describing the road network. Hence, GPS data typically needs to be aligned with the road network before further analysis can take place; this process is called map matching. The challenges in map matching are exacerbated when trajectories are sparse (have a low measurement frequency); we then need to fill the gaps between the measurements with a realistic estimate of the actual route between two points. As vehicle movement is subject to physical constraints (such as acceleration and speed bounds), an estimated route should be physically consistent. We present a method that creates physically consistent estimated routes to perform map matching for sparse trajectories.