Handling Conflicts Between Foot Gestures and Daily Foot Activities
Mobile and wearable devices are usually controlled with hands. However, humans often face situations where their hands are occupied with various activities like cooking and carrying grocery bags. Prior studies revealed that using foot gestures enables interaction with computers in scenarios where people cannot use their hands. While promising, it is unclear if these gestures will still be usable when used daily, since they have only been tested in a controlled setting. In this paper, we present a state-machine based gesture recognizer using atomic foot actions that detects the conflicts between the gestures and the regular activities. By getting atomic actions, our recognizer provides users an option to check potential false positives and fix the gestures based on their understanding of the situation. Our state-machine based recognizer outperforms state-of-the-art recognizer in terms of handling conflicts by reducing it. With a modification of a few conflicting gesture-set, we show that the false positive rate is reduced to <7% from 35%, which is approximately 80% reduction while keeping the true-positive rate intact.
- John A. Stankovic (Chair)
- Seongkook Heo (Advisor)
- Sebastian Elbaum
- Brad Campbell