The Self-driving Car Distribution Route Planning Based On ALNS

Self-driving technology is an important way to promote the transformation and upgrading of the transportation industry.When the traditional finished car is changed into a self-driving car that can be distributed on its own,it will also change the current finished car’s distribution situation of high distribution cost and distribution time.At present,the finished car distribution from the distribution center to the dealer needs to rely on the truck to carry out in the form of running ring line,which is of high cost and low efficiency.With the addition of self-driving car self-distribution,the composition of distribution route is extended to three forms: selfdriving car self-distribution,truck transport distribution,truck transport and self-driving car collaborative distribution,which brings more flexibility for distribution and provide the possibility for the system to reduce cost and increase efficiency.The self-driving car plays the dual role of goods and distribution resources in the selfdriving car distribution problem.Compared with the traditional distribution problem,the distribution network changes from multi-ring network structure to multi-ring and multi-branch network structure.As a result,most of the current heuristic algorithms of feasible solution construction methods based on the route which is a ring can’t be applied to this problem.The relationship between self-driving car distribution and truck distribution and the diversity of connection points between self-driving car distribution and truck distribution bring many difficult to solve the problem.Therefore,this paper tries to study the distribution route planning problem of the self-driving car.The research contents and innovations of this paper are as follows:(1)Aiming at the distribution route planning problem of the self-driving car,the distribution model of collaborative distribution between trucks and self-driving cars is constructed,the relevant constraints between self-driving cars and trucks are established and a certain time cost to weigh the paradoxical relationship between efficiency and cost is set.(2)The adaptive large neighborhood search(ALNS)is designed to solve the distribution route planning problem of the self-driving car,and the composition of the neighborhood is changed.Based on Removal heuristics and Insertion heuristics,Merger heuristics and Splitter heuristics are added to destroy and reconstruct the truck route and the self-driving car selfdistribution route respectively to promote the generation of the optimal solution.(3)The results of ALNS are compared with Gurobi solver,NST and memetic algorithm(MA),and 19 optimal solutions are obtained in 25 instances,with an average value of 0.52%.Then,the sensitivity analysis of the influence of the maximum distance of self-driving car self-distribution on the planning results is carried out,and it is concluded that with the increase of the maximum distribution distance of the self-driving car,it is more obvious to reduce costs and increase efficiency.It shows that with the increasing maturity of self-driving technology,self-distribution of self-driving cars will become an inevitable trend.This paper studies the distribution route planning of self-driving cars and establishes a cooperative distribution route planning model of self-driving cars and trucks considering distribution cost and distribution efficiency.Merger heuristics and Splitter heuristics are added to the ALNS to solve the problem,and a high-quality solution is obtained.This study not only introduces a new distribution mode but also provides an effective scheme for the implementation of self-driving car distribution route planning.