Project: #46 Multimodal Detection of Driver Distraction Progress Report - Reporting Period Ending: March 31, 2018 Principal Investigator: Maxine Eskenazi Status: Completed Start Date: Jan. 1, 2017 End Date: Aug. 31, 2018 Research Type: Advanced Grant Type: Research Grant Program: MAP-21 TSET National (2013 - 2018) Grant Cycle: 2017 TSET UTC Progress Report (Last Updated: March 20, 2018, 6:23 a.m.) % Project Completed to Date: 38 % Grant Award Expended: 38 % Match Expended & Document: 6 USDOT Requirements Accomplishments We have, as mentioned above, created the first level of features for the forward-looking information, the cognitive load and the CAN information. We are working on the backward-looking information at present. This first level of features represents the raw data that we receive and, for the second level, will need to be combined with other information. For example, steering wheel turning will be combined with forward looking curve vs non-curve detection; head turning can be combined with forward looking appearance of a sign, etc. Impacts We strongly believe that the algorithms we are developing will achieve better results than present in-car systems in detecting distraction due to both the massively multimodal nature of the data we capture and use and the way in which we implement the neural networks so that we leverage their predictive nature in detecting when an individual is distracted much earlier than present systems can. We also believe that our datasets, which will be shared with the research community, will help advance the field of distraction detection in general. Other Database of distracted driving using a simulator Software to detect driving distraction Outcomes New Partners none Issues It is not easy to get subjects to participate in the study. We have posted notices and sent out email. We find that a non-negligible portion of the people who contact us do not show up for their study appointment. We may ask our colleagues in the UTC to ask their students to participate.