Mining GPS Traces to Improve Digital Maps

Digital maps are an essential component of any automobile navigation system, but existing maps are both inaccurate and incomplete. However, the availability of accurate global positioning systems suggests an approach to improving them. This project focuses on collecting and storing time-tagged position traces from driven vehicles, then using these traces to discover useful knowledge about the road network and driving habits.

Applications to date include revising estimates for the centerline of a road segment, inferring the number of lanes on that segment, predicting a driver's travel time for the segment, and determining the presence and type of traffic controls at an intersection. Initial experiments results suggest that data mining of vehicle traces can substantially improve the accuracy and completeness of digital maps, and it clearly holds the potential to provide personalized models of driver behavior.

This work was supported by (and carried out at) the DaimlerChrysler Research and Technology Center. Researchers who contributed to the project have included Seth Rogers, Stefan Schroedl, Philip Tsai, Pat Langley, Simon Handley, and Huy Ton.

Related Papers

Schroedl, S., Wagstaff, K., Rogers, S., Langley, P., & Wilson, C. (2004). Mining GPS traces for map refinement. Knowledge Discovery and Data Mining, 9, 59-87.

Rogers, S., Langley, P., & Wilson, C. (1999). Mining GPS data to augment road models. Proceedings of the Fifth International Conference on Knowledge Discovery and Data Mining (pp. 104-113). San Diego, CA: ACM Press.

Handley, S., Langley, P., & Rauscher, F. A. (1998). Learning to predict the duration of an automobile trip. Proceedings of the Fourth International Conference on Knowledge Discovery and Data Mining (pp. 219-223). New York: AAAI Press.

For more information, send electronic mail to

© 1997 Institute for the Study of Learning and Expertise. All rights reserved.