Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : October 08, 2016 - October 10, 2016
It is very important for designers to ensure safety and usability of products. Since detecting problems of safety and usability through user observation manually may have some difficulty from the viewpoints of time, efforts and reliability, a computerized aid is expected. In this paper, computerized methods of (1) obtaining information of user operations on a product using RGB camera and depth sensor and (2) detecting potential problems of safety and usability. If designers make a checklist of possible problems (blacklist) to be detected, the list may contain only expected problems and thus unexpected problem could be missed. In the authors’ method, the authors define conditions on safety and usability about a target product (whitelist) and detect users’ operations that do not meet the conditions. With this whitelist-approach, unexpected problems can also be detected. The proposed method is applied to an interface of a tractor, and the results such as improved recall and precision of automatic user operation recognitions are reported.
