Research On Task Assignment Of Spatiotemporal Crowdsourcing Under Supply And Demand Uncertainty

Crowdsourced package delivery(CPD)is a new logistical mode which brings spatial crowdsourcing technology into logistics field.Task assignment is one of the core problems in CPD.However,the demand and supply are uncertain with location,time and price constraints varying,so how to balance the profit between the demand and supply to optimize the task assignment under the uncertain supply and demand conditions is particularly critical.On the basis of former research,this paper focuses on task assignment based on multi-objective optimization and dynamic pricing.The main research works are as follows:(1)Based on the constraints of maximum capacity and maximum waiting time window,the taxi-shared package delivery is extended to a multi-objective optimization problem.Furthermore,a framework named MOOP4 PD consisting of taxi pruning and multi-objective optimization process is proposed.In the solution process,MOOP4 PD uses NSGA-II algorithm to obtain the non-dominated solution set,and then uses the constraint solution method to find the ideal solution.Through the comparative analysis of simulation experiments,MOOP4 PD performs better than the baseline algorithm in terms of the success rate and cost of package delivery.(2)In order to solve the problem of profit-driven task assignment(PTA),a dynamic pricing algorithm based on Domain-of-influence(Do I)is proposed to address the limitation of fixed unit price of the traditional SC task assignment.In the task assignment stage,the time discount theory is applied to estimate the task acceptance probability of workers,and a greedy iterative refining algorithm is proposed to assign packages to the workers with high task acceptance probability,so as to solve the problem of supply and demand uncertainty in PTA.The results show that the proposed method can effectively improve the success rate of task assignment and the total revenue of the platform.(3)To deal with the CPD problem under uncertain supply and demand,a three-stage Crowdsourced Delivery Framework(CDF)is designed,and a crowdsourced package delivery algorithm namely Do IPrice is proposed to solve the CPD with the objective of minimizing delivery cost.Specifically,in the dynamic pricing stage,the proposed Do I-based pricing method considers the imbalance between supply and demand in local regions.In the package assignment stage,Do IPrice adopts three assignemnt strategies,i.e.,Minimum-price Maximum Assignemnt(MPMA),Minimum-time Maximum Assignment(MTMA)and Minimum Unitprice Maximum Assignment(MRMA).Finally,by comparing the experimental results with other classical methods,it is verified that Do IPrice can effectively reduce the package delivery cost.(4)Within the CDF,a crowdsourced package delivery algorithm DTPrice is also proposed to embody the Delaunay Triangulation(DT)to capture the dynamics of demand and supply,which avoids the iterative pricing calculation of Do IPrice and gains execution efficiency.In the package assignment stage,DTPrice adopts the optimal package assignment strategy(OPA)and greedy package assignment strategy(GPA)respectively.Among them,based on greedy assignment strategy,a new optimization assignment algorithm is proposed.The optimization process includes two steps,i.e.,package sorting and driver selection.The purpose is to give priority to the assignment of the high priced packages to the drivers with the highest psychological expectation reward.The experimental results show that DTPrice can effectively improve the total revenue of the platform.