Adaptive modality selection for navigation systems
Author(s)Kotowick, Kyle (Kyle Jordan)
Massachusetts Institute of Technology. Department of Aeronautics and Astronautics.
Julie A. Shah.
MetadataShow full item record
People working in extreme environments, where their mental and physical capabilities are taxed to the limit, need every possible advantage in order to safely and effectively perform their tasks. When these people -- such as soldiers in combat or first responders in disaster areas -- need to navigate through various areas in addition to performing other concurrent tasks, the combination can easily result in sensory or attentional overload and lead to major reductions in performance. Since the tasks that these people must perform often require intense visual attention, such as scanning an area for threats or targets, conventional visual navigation systems (map-based GPS displays) add to that visual workload and put users in danger of divided attention and failure to perform critical functions. This has lead to substantial research in the field of tactile navigation systems, which allow the user to navigate without needing to look at any display or use his or her hands to operate the system. While they have been shown to be extremely beneficial in many applications, tactile navigation systems are incapable of providing the detailed information that visual systems can and they make it more difficult to use the tactile sensory modality for other notifications or alerts due to tactile interference. This dissertation proposes a novel navigation system technology: one that adaptively and dynamically selects a navigation system's modality based on a variety of factors. Each modality has varying levels of compatibility with the different types of concurrent tasks, which forms the basis for the adaptive modality selection (AMS) algorithm. Additionally, there are time-varying factors called switching cost, sensory adaptation, and habituation that negatively affect navigation performance over long-duration navigation tasks; by switching between navigation system modalities when these effects have reached a point of notable performance loss, their effects can be mitigated. By considering both the task-specific benefits of each modality as well as the time-varying effects, an AMS navigation system can dynamically react to changes in the user's mission or environmental parameters to provide consistent, reliable navigation support. The research presented in this dissertation is divided into three phases, each involving a distinct human-participants experiment. The first phase investigated methods for selecting which modality to use for providing information to users when they are already completing other high-workload tasks. Results from the 45-participant experiment indicated that the primary consideration should be to avoid presenting multiple sources of information through the tactile modality simultaneously, suggesting that an AMS navigation system should ensure that the tactile modality is never used for navigation while it is also necessary for concurrent tasks. The second phase investigated the effects of sensory adaptation and habituation on navigation tasks, and evaluated whether it was possible to alleviate those effects by regularly changing between navigation system modalities. Results from the 32-participant experiment indicated that periodically changing between navigation system modalities induces a transient switching cost after each change, but that it also prevents long-term adaptation/habituation. The analysis indicated that the optimal time to change modalities was approximately once every five minutes. The third and final phase investigated the efficacy of an AMS navigation system algorithm, the design of which was informed by the results from the first two phases of research in combination with results from prior work. Participants were required to navigate while also performing various concurrent tasks while using a conventional single-modality navigation system, a multimodal system, or the novel adaptive system. Results from the 32-participant experiment indicated that when a user must both navigate and perform a concurrent non-navigation task simultaneously, use of an AMS navigation system can result in improved performance on both the navigation and the concurrent task.
Thesis: Ph. D. in Human Systems Integration, Massachusetts Institute of Technology, Department of Aeronautics and Astronautics, 2018.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Cataloged from student-submitted PDF version of thesis.Includes bibliographical references (pages 203-221).
DepartmentMassachusetts Institute of Technology. Department of Aeronautics and Astronautics.
Massachusetts Institute of Technology
Aeronautics and Astronautics.