Detecting Swipe Errors on Touchscreens Using Grip Modulation

摘要

We show that when users make errors on mobile devices they\udmake immediate and distinct physical responses that can be\udobserved with standard sensors. We used three standard cognitive\udtasks (Flanker, Stroop and SART) to induce errors from\ud20 participants. Using simple low-resolution capacitive touch\udsensors placed around a standard mobile device and the builtin\udaccelerometer, we demonstrate that errors can be predicted\udat low error rates from micro-adjustments to hand grip and\udmovement in the period shortly after swiping the touchscreen.\udSpecifically, when combining features derived from hand grip\udand movement we obtain a mean AUC of 0.96 (with false accept\udand reject rates both below 10%). Our results demonstrate\udthat hand grip and movement provide strong and low\udlatency evidence for mistakes. The ability to detect user errors\udin this way could be a valuable component in future interaction\udsystems, allowing interfaces to make it easier for users\udto correct erroneous inputs.