OBJECTIVE: Real-world driving studies, including those involving speeding alert devices and autonomous vehicles, can gauge an individual vehicle's speeding behavior by comparing measured speed with mapped speed zone data. However, there are complexities with developing and maintaining a database of mapped speed zones over a large geographic area that may lead to inaccuracies within the data set. When this approach is applied to large-scale real-world driving data or speeding alert device data to determine speeding behavior, these inaccuracies may result in invalid identification of speeding. We investigated speeding events based on service provider speed zone data.
METHODS: We compared service provider speed zone data (Speed Alert by Smart Car Technologies Pty Ltd., Ultimo, NSW, Australia) against a second set of speed zone data (Google Maps Application Programming Interface [API] mapped speed zones).
RESULTS: We found a systematic error in the zones where speed limits of 50-60 km/h, typical of local roads, were allocated to high-speed motorways, which produced false speed limits in the speed zone database. The result was detection of false-positive high-range speeding. Through comparison of the service provider speed zone data against a second set of speed zone data, we were able to identify and eliminate data most affected by this systematic error, thereby establishing a data set of speeding events with a high level of sensitivity (a true positive rate of 92% or 6,412/6,960).
CONCLUSIONS: Mapped speed zones can be a source of error in real-world driving when examining vehicle speed. We explored the types of inaccuracies found within speed zone data and recommend that a second set of speed zone data be utilized when investigating speeding behavior or developing mapped speed zone data to minimize inaccuracy in estimates of speeding.
BT - Traffic Inj Prev C1 - https://www.ncbi.nlm.nih.gov/pubmed/28379077?dopt=Abstract DO - 10.1080/15389588.2017.1315636 IS - 8 J2 - Traffic Inj Prev LA - eng N2 -OBJECTIVE: Real-world driving studies, including those involving speeding alert devices and autonomous vehicles, can gauge an individual vehicle's speeding behavior by comparing measured speed with mapped speed zone data. However, there are complexities with developing and maintaining a database of mapped speed zones over a large geographic area that may lead to inaccuracies within the data set. When this approach is applied to large-scale real-world driving data or speeding alert device data to determine speeding behavior, these inaccuracies may result in invalid identification of speeding. We investigated speeding events based on service provider speed zone data.
METHODS: We compared service provider speed zone data (Speed Alert by Smart Car Technologies Pty Ltd., Ultimo, NSW, Australia) against a second set of speed zone data (Google Maps Application Programming Interface [API] mapped speed zones).
RESULTS: We found a systematic error in the zones where speed limits of 50-60 km/h, typical of local roads, were allocated to high-speed motorways, which produced false speed limits in the speed zone database. The result was detection of false-positive high-range speeding. Through comparison of the service provider speed zone data against a second set of speed zone data, we were able to identify and eliminate data most affected by this systematic error, thereby establishing a data set of speeding events with a high level of sensitivity (a true positive rate of 92% or 6,412/6,960).
CONCLUSIONS: Mapped speed zones can be a source of error in real-world driving when examining vehicle speed. We explored the types of inaccuracies found within speed zone data and recommend that a second set of speed zone data be utilized when investigating speeding behavior or developing mapped speed zone data to minimize inaccuracy in estimates of speeding.
PY - 2017 SP - 845 EP - 851 T2 - Traffic Inj Prev TI - Perils of using speed zone data to assess real-world compliance to speed limits. VL - 18 SN - 1538-957X ER -