Jason Goldman, writing for Conservation:
The researchers started with a massive dataset: every trip taken by each of New York City’s 13,586 registered taxis that either started or ended in Manhattan in the year 2011. That gave them more than 150 million taxi trips. Each data point included the vehicle ID number, the GPS coordinates of the pickup and drop-off locations, and the travel time.
By passing that data through a graph-based mathematical model, the researchers identified opportunities for trip sharing without re-routing trips that had already started. The system that the researchers designed worked such that sharing options would have to be identified within one minute of the ride request. If no viable sharing options existed, then that request would initiate a new ride. That way, already-existing trips would not have to be re-routed; they would only pick up new passengers if the additional trip’s pickup and drop-off points were “on the way” to the original destination. By implementing such a program, the researchers estimate that transportation within Manhattan would become 40% more efficient.
The potential for this research is huge. Maybe not so much in North America where we seem to avoid interactions with strangers at all costs, but potentially in developing economies where vehicle-related congestion is just getting started.