Efficient Multi-modal Route Planning with Transfer Patterns

In these days, there is a big interest in route planning systems with different types of transportation. We study a way to efficiently compute optimal paths with real data in a multi-modal scenario considering walking, transit and car. Therefore, we implement Transfer Patterns, which is a state-of-the-art route planning approach [8]. This allows to obtain shortest paths queries within milliseconds. When using a Pareto-cost model various optimal paths are provided for the user. However, the high number of optimal solutions also includes a lot of unreasonable routes, although they are Pareto-optimal. To address this problem, we use a filtering procedure called Types aNd Thresholds (TNT) [10], which provides a both small and diverse set of reasonable optimal paths decreasing the number of transfer patterns and hence resulting in faster query times. To our knowledge this is the first work which realizes the combination of these two approaches (Transfer Patterns with TNT). We present an evaluation on query times of selected optimal paths, which shows different and reasonable route options with an average query time of 14 milliseconds for the Freiburg network.