Addressing the Need for Map-Matching Speed: Localizing Globalb Curve-Matching Algorithms

摘要

With vehicle tracking data becoming an important sensor data resource for a range of applications related to traffic assessment and prediction, fast and accurate mapmatching algorithms become a necessary means to ultimately utilize this data. This work proposes a fast mapmatching algorithm which exploits tracking data error estimates in a provably correct way and offers a quality guarantee for the computed result trajectory. A new model for the map-matching task is introduced which takes tracking error estimates into account. The proposed Adaptive Clipping algorithm (i) provably solves this map-matching task and (ii) utilizes the weak Fr´echet distance to measure similarity between curves. The algorithm uses the error estimates in the trajectory data to reduce the search space (error-aware pruning), while offering the quality guarantee of finding a curve which minimizes the weak Fr´echet distance to the vehicle trajectory among all possible curves in the road network. Moreover, this work introduces an outputsensitive variant of an existing weak Fr´echet map-matching algorithm, which is also employed in the Adaptive Clipping algorithm. Output-sensitiveness paired with error-aware pruning makes Adaptive Clipping the first map-matching algorithm that provably solves a well-defined map-matching task. An experimental evaluation establishes further that Adaptive Clipping is also in a practical setting a fast algorithm that at the same time produces high-quality matching results.