Department of Computer Engineering
CS 590 SEMINAR
Hierarchical Embedded Platform Development for Autonomous Driving
Ege Berkay Gülcan
Computer Engineering Department
In order to tackle challenges of autonomous driving (AD), increasingly complex algorithms are applied to AD systems in the form of artificial intelligence (AI) systems such as deep neural networks (DNN). However, these algorithms require powerful computational resources in the form of GPUs, multi-core CPUs and FPGAs, which is not suitable for embedded systems used in autonomous vehicles. To decrease the computational load of such systems, we propose an hierarchical heterogenous framework for autonomous vehicles. This framework consists of three platforms, namely; edge, fog and cloud. Our frameworks provides platform and model selections using data fusion techniques.
DATE: 02 December 2019, Monday @ 17:10