In many Western countries, governments are currently implementing an innovative demand-driven mobility policy. Providers of collective door-to-door transport, called dial-a-ride services, are increasingly invoked to replace unprofitable public transport in rural areas. This requires an integrated mobility system in which a user's trip may consist of a combination of dial-a-ride services and regular public transport.

In order to optimally integrate both systems from an operational point of view, dial-a-ride providers need to solve a challenging routing problem which integrates the opportunity to use public transport in the classical dial-a-ride problem. Dial-a-ride routes should be synchronized to the timetables of the public transport services, while the optimal selection of the users' transfer terminals depends on the actual structure of the dial-a-ride routes.

This paper introduces a metaheuristic routing algorithm, based on Large Neighbourhood Search, to solve this integrated routing problem. An exact scheduling procedure is embedded to enforce the synchronization between dial-a-ride routes and public transport.

Experiments, performed on a new artificial benchmark data set with realistic characteristics, clearly indicate that considerable operational benefits are obtained by integrating dial-a-ride services and public transport. The resulting distance savings for the dial-a-ride vehicles are shown to depend on the operational characteristics of the system, the geographical distribution of the demand, and the ability to flexibly assign transfer terminals to user requests.