Fishing

Abstract

An efficient selection protocol is needed in an IoT environment with many smart objects. As the number of objects increases, unintended selection can be made more often, and selecting a target among multiple objects can be more challenging. Resolving these issues has often required precise body tracking or spatial cursor visualization, which are not practical in a smart home context. Therefore, we propose Fishing, a selection protocol where the user selects an object through a bidirectional time-window-based interaction. We showed that Fishing could provide a stable selection performance with a high selection speed (1.79-3.32s) and accuracy (93.8-98.8%). Fishing's accuracy remained high independent of the object density and yielded 2.4 times fewer unintended selections from ambient indoor hand movements than a state-of-the-art disambiguation protocol. Despite its bidirectional protocol structure, the Fishing selection time was comparable to the simulated selection time of the natural hand-pointing model.