We have two important issues in the field of railway facilities maintenance. One is because of a decrease of experienced maintenance staff due to the generational change, it is required to save labor for facilities inspection data input. The other is that we have to substantially hand over technical information owned by experienced staff to the next generation. In this research, for the former issue, we examined a hands-free data input system using speech recognition technology, and for the latter issue, we developed a database of technical documents for facilities maintenance manuals and constructed a prototype system for offering necessary information using a mobile device with spoken dialogue expressed in natural language. We have adopted XML to describe facilities maintenance manual documents, because XML makes it possible to naturally express their hierarchical structures and to add or remove the data easily. This paper reports development of the two systems and the results of their evaluation tests in the maintenance fields and also discusses the future related works.

Copyright (c) 2006 Railway Technical Research Institute