- 1 - A Fast Scheduler for Multiagents in a Warehouse Jose Ildefonso U. Rubrico*, Toshimitsu Higashi**, Hirofumi Tamura***, Makoto Nikaido*, and Jun Ota* *Dept. of Precision Engineering, The University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo, Japan E-mail: ota@robot.t.u-tokyo.ac.jp **Murata Machinery, Ltd., 2 Nakajima, Hashizume, Inuyama-shi, Aichi, Japan ***Murata Systems, Ltd., 3 Minamiochiai-cho, Kisshoin, Minami-ku, Kyoto, Japan Abstract The problem of scheduling multiagents for picking products in a warehouse is considered. The goal is to decrease operating costs by minimizing the maximal makespan among transport agents. Computational time must be kept within ten seconds even for real-sized instances. The orders are initially batched by solving a Split Delivery Vehicle Routing Problem. The resulting trips are then assigned to the agents such that their picking times are balanced. Finally, the trip assignments are improved to minimize the blocking delays of the agents. The simulation results show that the proposed method improves the existing method. Keywords: scheduling, planning, multiagents, automation, Split Delivery Vehicle Routing Problem 1. Introduction Warehouse management has been gaining popularity over the years due to significant benefits that can result in operational efficiency. Warehouse management problems are categorized into planning issues and control problems [1]. The former are concerned with inventory management and storage location assignment [2], and the latter deal with storage and retrieval operations [3]. Studies have shown that order picking (retrieval) dominates the activities in modern warehouses [4]. Thus, special consideration is given to optimizing the picking operations at a warehouse. Traditionally, the Warehouse Picking Problem (WPP) has been considered to be a distance minimization problem in which items for an order are manually picked by a single agent traversing narrow aisles. It is represented as a form of the Traveling Salesman Problem (TSP) and then solved