[4Q3-OS-14] AI for Scienceにおける再現性と信頼性

[4Q3-OS-14-03] Self-maintainable science laboratories

〇Koji Ochiai¹, Kazunari Kaizu¹, Koichi Takahashi^1,2,3 (1. RIKEN, 2. Graduate School of Media and Governance, Keio University, 3. Graduate School of Frontier Biosciences, Osaka University)

Keywords:robot, Automated laboratory

While the automation of scientific experiments brings many benefits such as improved reproducibility, efficiency, and accuracy of experiments, there is a new need for keeping the automated system itself functioning properly and always maintained. In this paper, we coin the term self-maintainability as the ability of a system to maintain itself in a state that allows it to perform its functions in an environment that involves the consumption of supplies, wear and tear on equipment, and the possibility of operation failures due to disturbances. We examine its potential benefits and challenges. Self-maintainability could significantly improve reliability, uptime, and cost-effectiveness of systems, and thus be a useful property not only for automated laboratories but also for future production and urban infrastructure systems.