Customised Shortest Paths Using a Distributed Reverse Oracle

Abstract

We consider the design and implementation of a centralised oracle that provides commuters with customised and congestion-aware driving directions. Computing directions for a single journey is straightforward, but doing so at city-scale, in real-time, and under changing conditions is extremely challenging. In this work we describe a new type of centralised oracle which combines fast database-driven path planning with a query management system that distributes work across a small commodity cluster of networked machines. Our system allows large-scale changes to the underlying graph metric, from one query to the next, and it supports a variety of query types including optimal, bounded suboptimal, time-budgeted and k-prefix. Simulated experiments show strong results: we can provide real-time routing for all peak-hour commuter trips in the city of Melbourne, Australia.

Snapshot-Optimal Real-Time Ride Sharing

“…The road network in Manhattan, New York City consists of 4484 nodes and 8839 edges with an average edge distance of 186 m. Unlike previous studies that employed designated pickup/drop-off points [11], SnapPair considers every node for pickup and dropoff. For edge costs, we compute free-flow travel times in seconds by dividing edge length by the maximum allowed speed, following similar methods in other research [38]. Our travel demand consisted of 23,981 requests covering the hour-long morning peak from 8-9 a.m. on 1 May 2013, a working Wednesday.…”

Reducing Redundant Work in Jump Point Search

“…CCH is fast, optimal and it does not rely on any cost-based assumptions. A main disadvantage is that as the number and frequency of map changes increases the amortized cost of repair grows large, to the point where it becomes faster to use a reference algorithm such as A* (Mahéo et al 2021).…”

“…The reason is that preprocessing-based approaches rely on precomputed auxiliary data structures, which have to be rebuilt or repaired when the map changes. When dynamic changes are frequent and/or affect large regions of the map, the online costs of rebuild or repair operations grows quickly and the amortized performance of preprocessing-based algorithms becomes worse than purely online approaches (Mahéo et al 2021;Hechenberger et al 2020).…”