Scalable space-time trajectory cube for path-finding: A study using big taxi trajectory data

Abstract

Route planning is an important daily activity and has been intensively studied owing to their broad applications. Extracting the driving experience of taxi drivers to learn about the best routes and to support dynamic route planning can greatly help both end users and governments to ease traffic problems. Travel frequency representing the popularity of different road segments plays an important role in experience-based path-finding models and route computation. However, global frequency used in previous studies does not take into account the dynamic space-time characteristics of origins and destinations and the detailed travel frequency in different directions on the same road segment. This paper presents the space-time trajectory cube as a framework for dividing and organizing the trajectory space in terms of three dimensions (origin, destination, and time). After that, space-time trajectory cube computation and origin-destination constrained experience extraction methods are proposed to extract the fine-grained experience of taxi drivers based on a dataset of real taxi trajectories. Finally, space-time constrained graph was generated by merging drivers’ experience with the road network to compute optimal routes. The framework and methods were implemented using a taxi trajectory dataset from Shenzhen, China. The results show that the proposed methods effectively extracted the driving experience of the taxi drivers and the entailed trade-off between route length and travel time for routes with high trajectory coverage. They also indicate that road segment global frequency is not appropriate for representing driving experience in route planning models. These results are important for future research on route planning or path finding methods and their applications in navigation systems.