Research On The Dynamic Scheduling Problem Of Meal Delivery In The Joint Service Of Riders And Unmanned Vehicles

With the development of the Internet and the rise of the "home economy",the meal delivery industry in China has developed rapidly.To advocate "contactless delivery" services,major meal delivery platforms have successively launched new models of unmanned delivery.Moreover,there is a clear trend of consumption upgrading,people need not only ordinary meals,but also some perishable food that must be kept at low or high temperatures,such as ice cream,hot coffee,and strawberry.Unmanned delivery vehicles with temperature control can meet these needs mentioned above,and riders can complement unmanned delivery vehicles to provide door-to-door service.Therefore,this paper studies the dynamic delivery scheduling problem with the mixed fleet of riders and unmanned delivery vehicles.The main works are as follows:(1)The meal delivery mode of riders and unmanned delivery vehicles is proposed.Consider meal delivery modes in China and the development trend,a meal delivery mode based on a mixed fleet of riders and unmanned delivery vehicles is proposed.The characteristics of the mode are analyzed,and the business process is designed.The autonomous vehicle is a pure electric vehicle.There are multiple meal locks in the smart container.Each lock can independently control the temperature.When the vehicle’s power is insufficient,the battery can be replaced at the battery swap stations.Adapted vehicle types may be provided based on customer preference or venue characteristics.(2)A dynamic meal delivery scheduling problem is proposed and a mathematical model is established.Considering the constraints such as order of delivery,vehicle type,capacity,and electricity,a dynamic Pickup and Delivery Problem with Mixed Fleet and soft time windows(DPDPMF)is proposed.A mixed integer programming model for DPDPMF is developed to minimize vehicle fixed cost,rider delivery fee,energy consumption cost of unmanned delivery vehicles,and penalty cost.(3)The Adaptive Large Neighborhood Search heuristic(ALNS)is designed.According to the characteristics of DPDPMF,19 operators are designed,such as Intermittent Request Removal(IRR),Same Restaurant Removal(SRR),Same Customer Removal(SCR)and so on.The adaptive weight adjustment strategy and the simulated annealing acceptance criterion are adopted,so that the neighborhood with better performance in the previous iterations will be selected in the subsequent iterations with a higher probability.(4)Experimental verification.A benchmark instance set of DPDPMF is constructed,and parameters are set through parameter tuning experiments.The performance of the ALNS is tested by using instances with different scales to verify the effectiveness of the method.Finally,ALNS is applied to solve large-scale instances.Handling dynamic orders through the Receding Horizon Control.Indicating the usefulness of the ALNS.