Crowdsourcing Logistics Study Based On Taxi Trajectory Data Mining

With the rapid development of the Internet and the pervasive use of tablet and mobile devices,the round-the-clock online shopping has become more and more popular.In order to meet people's increasing demand for online shopping,logistics industry is developing sustainably.However,it is hard for tradition package delivery to solve the arriving-on-time package delivery of the same city,the low-cost demand for mass parcel delivery,the resulting urban traffic congestion etc.As the new model of parcel delivery,crowd logistics dispatch the task of parcel delivery to the crowd,which fully exploits the spare carrying capacity of society to solve the problems existing in the tradition package delivery.Firstly,this article proposes a new model of crowd logistics to parcel delivery,which hitchhikes the urban taxis that are transporting passenger to relay packages collaboratively.Then based on the taxi trajectory mining and the characteristics of crowd logistics,the package transportation network was established.Based on the network,an offline parcel delivery model with the goal of minimizing the extra travel distance and an online real-time parcel delivery model with the goal of maximum arriving-on-time probability of parcel delivery are established.The former is manly to build an integer linear optimization problem with the goal of minimizing the extra travel distances,which is to solve the matching problem of parcel delivery tasks and taxis statically.That is to find the optimal match of delivery tasks and taxis with the parcel delivery requests and passenger-delivery trip known ahead.The latter proposes a dynamic adaptive taxi scheduling algorithm to solve the stochastic optimization problem based on the real-time parcel delivery request for parcel delivery routing with the goal of maximizing the arriving-on-time probability.That is to find parcel delivery routing and schedule taxi by the passenger-delivery trip,which comes with time as in the realistic.Finally,model system experiments with history taxi passenger trajectory data in New York City.Experimental results show that the offline parcel delivery model generates an average travel distance of about 500 meters per parcel relay.The online real-time parcel delivery model was able to complete parcel deliveries of approximately 9,500 with the arrivingon-time success rate over 94% during the daytime,and its average computation time was within 25 milliseconds.It shows that this parcel delivery model can assist the delivery of massive parcels in the city at a lower extra cost,complete the parcel delivery on time with high success rate and be able to respond the emerging request quickly.