Mid-air gestures for manipulation of multiple targets in the physical space - comparing the usability of two interaction models.

The design of mid-air, accessory-free, gesture-based control of multiple targets (devices) in the physical space presents several challenges, including: (a) user control entails both addressing the device and providing input to it (either sequentially or simultaneously); (b) despite a potentially large number of devices, the gesture vocabulary must be kept small to allow for memorability; (c) the gesture set must include discrete gestures to avoid the phenomenon of Midas’ touch (accidental operations or false positive errors). We have addressed those challenges by introducing two interaction models: (a) the two-stage approach in which the user applies sequentially separate gestures for Address and then Command a device (A-t-C model), and (b) one-stage approach in which the user applies a two-handed gesture that simultaneously Addresses and Commands a device (A & C model). In this paper, we first describe the two interaction models and then we present their comparative usability evaluation. We have found that both interaction models present very high task success and very low false-positive and memorability errors. Furthermore, the A&C model is significantly more usable than A-t-C in task-time and false-negative errors (i.e. errors of no system response). These results are very encouraging for future applications of mid-air gesture control of multiple devices in the physical space.