Unterschiede

Hier werden die Unterschiede zwischen zwei Versionen angezeigt.

Link zu dieser Vergleichsansicht

Beide Seiten der vorigen RevisionVorhergehende Überarbeitung
lehre:ws18:fsm_18ws:group_e [21.03.2019 10:29] – Updated short description ost03246lehre:ws18:fsm_18ws:group_e [22.03.2019 22:42] (aktuell) – Updated short description ost03246
Zeile 5: Zeile 5:
 members_            : Paul Winderl, Johannes Dengler, Thomas Oswald members_            : Paul Winderl, Johannes Dengler, Thomas Oswald
 keywords_           : WIP, measuring, user study, input device, latency keywords_           : WIP, measuring, user study, input device, latency
-shortdescription    : Latency measurement is no new research area but has ongoing relevance to users of input devices. As this number is not decreasing, we investigate a new approach. Bockes, Schmid and Wimmer [1] described an invasive way to measure input latency of different types of devices. We pick up their idea by doing the same with a non-invasive approach. Which means, that we focus on pressing a button of USB-connected input-devices to trigger an input-event and measure the time it takes to arrive at the system. +shortdescription    : Research in latency measurement is conducted since decades, but has ongoing relevance to users of input devices. Therefore we investigate a new approach to this topic. Bockes, Schmid and Wimmer [1] describe an invasive way to measure input latency for different types of devices. We pick up their idea and conduct experiments with the same devices but using a non-invasive approach. This means, that we focus on pressing a button of an input-device that is connected to a Raspberry Pi 2 via USB to trigger an input-event and measure the time it takes to arrive at the system. The button-press is achieved by a robot that physically triggers said button through the movement of a measuring head. 
-To obtain the moment of the click-event we capture the timestamp immediately after the stepper motor moved down a beforehand calibrated distance. A second timestamp is saved when the USB-event arrives. Both values subtracted return our calculated input latency. +To obtain the exact moment of the click-event, we capture the timestamp immediately after the stepper motor moved down a definite distance, which is calibrated beforehand. A second timestamp is saved when the USB-event arrives. Subtracting the first timestamp from the latter returns our calculated input latency. 
-This approach will be helpful in research and industrial areas to assess the latency of input devices. We implemented a reliable process to measure latency for a selection of devices and got close to the findings of the invasive approach. However, with our current method, we can't robustly assess all device types with buttons. Further, studies will need to use other means, such as an algorithm for the pressure curve, to reach that goal.+This approach will be helpful in research and industrial areas to assess the latency of input devices. We implemented a reliable process to measure latency for a selection of devices and got close to the findings of the invasive approach. However, with our current method, we can't robustly assess all device types. Further, studies will need to use other means, such as an algorithm using differential geometry for the pressure curve, to reach that goal.
 ---- ----
 A non-invasive approach for measuring input device latency: Method for input device latency measurement without irreversibly damaging the device. A non-invasive approach for measuring input device latency: Method for input device latency measurement without irreversibly damaging the device.