Dynamic vehicle routing problem analysis for AGV handling multiple part types

Abstract

In this dissertation, the validity of the vehicle system handling multiple part types is to be verified, and the policy is to be advanced. In the circumstance where vehicle system handling only a single part type is generally used, this study is conducted from discussions on the need for a vehicle system handling multiple part types. The problem situation is approached as a vehicle routing problem, and dynamic vehicle routing problem with multiple types of demand that the demands occur in time according to a Poisson process is studied. The objective of this study is to minimize the sum of waiting time of demands. In addition, studies about vehicle routing problems so far generally address only one type of demand, but this study addresses the problem with multiple types of demand. Therefore, the vehicle scheduling policy should be established in consideration of the knapsack problem, one of the combinatorial optimization problems. In this dissertation, the decision-making process of the dynamic vehicle routing problem is formulated by the Markov decision process (MDP). Based on the formulated model, dynamic programming (DP) determines vehicle scheduling policy, and the policies obtained are analyzed to provide insights for solving this problem.