The Effect of Latency on High Precision Micro Instructions in Mobile AR

The effectiveness of AR superimposed instructions with video see-through displays has been extensively explored due to the widespread availability of suitable handheld devices such as smartphones. However, looking through a camera lens in such devices affects users’ performance due to distortions in visual representation and hardware capacity. While the former have been studied in detail (distortion of binocular disparity, accommodation/vergence conflict, geometrical distortions), the latter, mainly affected by system latency, has not been explored much. In this work, we present a system to measure the latency and two user studies: one exploring how well users can perceive differences in system latency, and one exploring the effects of latency on completion time and accuracy in following high precision micro AR instructions. The results show that users start to notice the difference in two compared latencies at 69 ms and that latency did not affect the performance (times and accuracy were similar in all conditions). Nevertheless, when ranking the phones with different latencies users preferred the phones of latency values 132 ms or lower. Both results provide a guide for building the best possible human perceivable system for a given task.