The rise of ubiquitous computing has led to the rapid development of intelligent systems such as augmented reality (AR). AR is being employed for various usage including common tasks such as web browsing, navigating and controlling other internet of things (IoT) devices. In these types of system usage, similar to mobile phones, the presentation of notifications is inevitable. However, studies with regards to the effective delivery of notification signals within the AR environment is still lacking. Thus, the present study aims to investigate the effects of different output modalities (visual, auditory and multimodal) for presenting notifications during simple search and selection tasks within the AR system. Twenty individuals participated in a user study where they were asked to perform a simple search and selection task while using the AR device. Participants were asked to confirm the notifications that were presented at random in each trial. Results of the user study showed that multimodal notification signals (both visual and auditory) yielded a higher notification recognition rate and were rated as the most appropriate modality and most preferred among the conditions tested. Also, it was found that speech input modality was used more frequently in response to notifications, regardless of the output type. Several design implications were derived from the findings of the study.

Interruptions while driving can be quite dangerous, whether these are self-interruptions or external interruptions. They increase driver workload and reduce performance on the primary driving task. Being able to identify when a driver is interruptible is critical for building systems that can mediate these interruptions. In this paper, we collect sensor and human-annotated data from 15 drivers, including vehicle motion, traffic states, physiological responses and driver motion. We demonstrate that this data can be used to build a machine learning classifier that can determine interruptibility every second with a 94% accuracy. We present both population and individual models and discuss the features that contribute to the high performance of this system. Such a classifier can be used to build systems that mediate when drivers use technology to self-interrupt and when drivers are interrupted by technology.