Region-Aware Route Planning

We consider route planning queries in road or path networks which involve a user preference expressed in relation to a spatial region, as e.g. 'from Nanjing to Shanghai along Yangtze river' or 'from home to work through Central Park'. To answer such queries, we carefully define relevant subgraphs of the network for each region-of-interest and guide the route towards them. To extract these subgraphs, we need to solve several non-trivial geometric problems (as computing weak visibility regions), which require to interpret the embedded network both as a graph and as an arrangement of line segments. We describe a suitable preprocessing framework, taking the special structure of road networks into account to increase its performance. Our query answering algorithm then allows to trade detour length against time spent within or close to the desired region. Using acceleration techniques, region-aware routes can be planned efficiently even in networks with millions of edges, and also when considering large or complex regions.