Driver behavior monitoring

A significant number of road fatalities could be averted if a driver’s overt/covert behavior is monitored in real time. Overt abnormal behavior such as that related to drowsiness (or driving in an intoxicated state) can be detected by a camera with the ability to analyze video feeds locally at a DG installed in the vehicle. Trained ML models for analyzing video feeds are pushed periodically from a central server. These models gauge and report driver behavior, along with a confidence level. Analysis of behavioral patterns can be further augmented by obtaining additional data from the driver’s wearable device (such as pulse rate and last night’s sleep quality).

Processing needs to be done locally (at the edge) as video feeds can’t be sent to the central server as it would hog the communication channel’s bandwidth, as well as the response not being within expected time limits.

Along with the video feed, other crucial...