50. DeepROAD: A Multifaceted Deep Learning Suite for Real-Time Optimized Autonomous Driving
Authors: Edwin L. Weill (Clemson University)Jesse Tetreault (Clemson University)Varun Praveen (Clemson University)Melissa Smith (Clemson University)
Abstract: Autonomous driving is undergoing intensive study in various fields of research. The number of research topics directly related to the success of an effective autonomous vehicle is proliferating. Online assessment of surroundings, one of the most important actions for an autonomous vehicle, has had a large number of research projects attempting to tackle different aspects of driving; for instance, there is research geared toward control of the vehicle while other research is focused on understanding the environment. DeepROAD intends to leverage deep learning in a manner conducive to perception and understanding of surroundings in real-time. This research employs deep learning for detection and segmentation of surroundings for decision making as well as compression, allowing for smaller networks and quicker inference times.
Two-page extended abstract: pdf