Urban passenger transit is multimodal in nature, and connectivity at the last-mile with the dominant public transit system in a city is critical. In this paper we develop and demonstrate a replicable framework that comprehensively defines a last-mile universe by quantifying coverage, identifying potential last-mile stops and the associated last-mile feeder network for any urban geography where the public transit network is known and associated spatio-demographic information is available. The methodology is based on spatial decomposition techniques from computational geometry using principles of tessellations, Voronoi diagrams, and Delaunay triangulations from computational geometry.

We demonstrate the proposed framework utilizing publicly available data for an urban agglomeration in India (Bengaluru) over its planned metro rail infrastructure. We find that the proposed long-term metro rail network covers about 43% of the current population of 8.4 million within a walking distance of 1 km from the nearest station, spread over 32% of the urban area. The methodology identifies 245 last-mile stops and sub-zones within the administrative region and recommends a feeder network structure serving 92 stations. These can be used to deploy last-mile feeder systems which can potentially bring 93% of the city’s population under walkable coverage of a multimodal metro plus last-mile setup, spanning 89% of the area.