Enhancing Driver Awareness at the Edge

From calls, texts, e-mails, and even mobile games, distracted pedestrians and drivers are prevalent on roadways today. The National Highway Traffic Safety Administration’s National Center for Statistics and Analysis published a report of statistical analysis related to distracted driving. It was reported that about 10% of vehicle crashes as well as 10% of fatalities reported were distraction-affected. Until the time comes when crash fatalities from distracted driving can be completely eliminated, solutions must be found to help lessen the risks. Imagine if there was a way to have an extra two to three seconds added to the time between a driver being made aware of an obstacle and having that extra time to react to the situation. Or imagine if any two objects’ predicted trajectories intersect a time ‘til collision value is calculated in real time and having this value can identify near miss cases to avoid collision. In this talk, we present our research on video analytics in real time and how it can be used to enhance driver and pedestrian awareness. Real-time video analysis at the edge will enable easily acquired hardware to play a more critical role than simply active monitoring. Machine learning and tracking algorithms are used to gather metrics on traffic flow, velocity, road blockages/closures. These metrics can be used to improve the driver and pedestrian awareness by providing a holistic view of the traffic and gaining enough knowledge about interactions in the surrounding are for both drivers and pedestrians. Additionally, we present our novel spectrum sensing algorithm executed at the edge that supports massive and real-time V2X communications needed to facilitate the information exchange between vehicles, pedestrians, and infrastructure.

Location: Cumberland Amphitheatre Date: August 29, 2019 Time: 9:45 am - 10:15 am Mina Sartipi (UT-C)