[1K4-OS-15a] OS-15

[1K4-OS-15a-04] Progress and Prospects of Simulation and AI Technologies in Rail Transport Planning

〇Shumpei Kubosawa^1,2, Takashi Onishi^1,2, Yoshimasa Tsuruoka^1,3 (1. AIST, 2. NEC, 3. Univ. of Tokyo)

Keywords:transport planning, reinforcement learning, simulation

There is a long history of research into automation techniques for rail transport planning. These are computer-based techniques for optimal planning from a given point of view, and some of them use simulation technology. However, the ultimate goal of these technologies is not to develop an optimal plan in a simulation, but to reflect the developed plan in real rail traffic and achieve better operations in the real world. This includes technologies to capture the detailed real world track conditions, user interface technologies and system integration technologies to interface with operations management systems. However, despite many previous research efforts, their practical application has long been difficult due to the complexity and scale of the problem. Therefore, this paper focuses on simulation and AI-based technologies and peripheral technologies related to rail transport planning, and discusses the progress made so far and future prospects for practical application.