Dynamic Control of Task Demands in Mobile Contexts using Sensor Data and Adaptive User Models

点击翻译按钮获取中文摘要

This research program will develop a theory and method for dynamic control of workload in mobile multi-task settings, encompassing mobile, wearable, and vehicular interactions with applications. How can applications be made aware of current workload demands and task-context so that they can reconfigure themselves dynamically to optimize safety and productivity in the current context for a specific user profile? This research will answer this question by developing new types of adaptive interface, informed by pattern analysis of sensor data and user models developed using statistical and machine learning techniques. Relevant use cases include a driver interacting with in-vehicle technology while driving, someone using a mobile application while navigating obstacles in a mine or a construction site, a pilot operating a light aircraft and carrying out multiple tasks in general aviation, or a person using a hand-held or wearable device and walking, while receiving messages and navigation instructions from friends and a map application.This research will develop an overall model for adaptive interfaces in mobile, wearable, and vehicular contexts where adaptation is based on dynamic control of key parameters of the primary task. Along with the overall model, we will develop specialized models for adaptive interfaces in selected use cases. The research will identify regions of overload and inattention in task parameter spaces that signal a need for adaptation in different task contexts, developing task-specific indicators of adaptation triggers such as drowsiness, distraction, or general overload. Mining and machine learning within relevant data repositories (e.g., driver interactions in instrumented vehicles, user interactions with mobile applications that include built-in gait analysis, log data from instrumented aircraft or flight simulators) will be used for dynamic user profiling and personalization of the adaptive interface in different contexts. This work is novel because general adaptive interfaces that protect the primary task from the interfering effect of a mobile, wearable, or in-vehicle application do not yet exist. While some adaptive interfaces have been developed for driving tasks, they are limited in scope and have not been generalized to a range of mobile contexts. I intend to develop a strong scientific model of adaptive interfaces that can be specialized for different mobile use contexts. This work will be highly significant for the design of applications intended for multi-task use in mobile settings. The results of this research will support the development of cutting edge applications such as safe driver notification systems, falls risk assessment for elderly users based on disruptions to gait caused by smartphone use, safety enhancement of mobile applications used in industrial settings, and improved interfaces for general aviation pilots.