表題番号：2025C-519
日付：2026/03/30
研究課題福祉ロボットの安全情報共有に関する研究
| 研究者所属（当時） | 資格 | 氏名 | |
|---|---|---|---|
| （代表者） | 理工学術院 大学院基幹理工学研究科 | 教授 | 本間 敬子 |
- 研究成果概要
- In order to ensure the safe use of assistive devices including assistive robots, it is important for stakeholders to properly share information regarding the safety of assistive devices. Therefore, this study focused on developing a tool that uses generative AI to make safety information regarding assistive devices including robots easier for users to understand, as a means of appropriately sharing safety information among stakeholders. A prototype of the tool was constructed using existing language models. The prototype was then used to generate explanatory texts on safety terminology. The following four conditions were established for the text generation process: the presence or absence of text simplification, and the presence or absence of the retrieval-augmented generation (RAG) technique. The evaluation of the generated explanatory texts was conducted based on three criteria: (1) difficulty, (2) comprehensiveness, and (3) accuracy. The evaluation results demonstrated that there was no significant change in difficulty due to text simplification. Furthermore, the implementation of RAG has been shown to result in marginal enhancements in comprehensiveness and accuracy. In future developments, the refinement of the prototype will be pursued to create a more practical tool, and the proposal of new research directions regarding the sharing of safety information will be undertaken.