Hitachi, Ltd. has agreed with East Japan Railway Company to conduct a joint trial starting around September 2025 to measure the effectiveness of AI agents in railway traffic management and maintenance operations in the Tokyo metropolitan area.
The Tokyo Metropolitan Transport Control System (ATOS), which manages the operation of conventional lines in the Tokyo metropolitan area, is a large-scale system that combines many devices in a complex manner. Therefore, when a problem or inquiry about the function occurs, the dispatcher who analyzes it and identifies the cause is required to have very high level of specialized knowledge and know-how. If it is difficult to solve the problem using the manual, it is necessary to contact an experienced person, and it may take time from identifying the cause of the failure to recover.
As the working population continues to decline and a shortage of dispatchers is expected in the future, the two companies will utilize AI agents as part of business transformation through digital transformation in order to achieve more stable transportation with fewer staff, with the aim of realizing safe and sustainable railway operations.
こちらもお読みください： 青葉警察、イッツコム、FMサルースが安全協定を締結
Specifically, Hitachi will use the OT (control and operation technology) knowledge it has acquired from hundreds of cases it has worked on, as well as the Lumada approach, which transforms data and knowledge into value, to build a large-scale language model (LLM) specialized for railway traffic management by incorporating the knowledge assets of both companies, such as JR East’s system specifications (explicit knowledge), 日立‘s Infrastructure Control Systems Division’s control device documents (explicit knowledge), and operational know-how (tacit knowledge). In addition, an AI agent based on a failure response scenario that reproduces the thought process of an expert will be developed, and by combining it with the LLM, it will be verified whether it can automatically identify the location of the failure and propose a response policy to support the decision-making of the dispatcher. This verification, which applies generative AI to a railway traffic management system, is the first attempt for both companies. Through these efforts, we aim to contribute to improving the efficiency and stability of traffic management operations.
ソース PRタイムズ
