ABSTRACT
Autonomous vehicle (AV) is described as a software-defined vehicle which requires robust hardware and software infrastructure to work between intensive resources tasks and real-time tasks. The use of advanced analytics techniques for this research requires two hardware/software systems: one for training in the cloud and another for the AV. This research looks at how the software architecture inside the AV distributes this type of algorithm, and how Deep Learning algorithms to detect traffic signals integrate in the system. The main findings are that Deep Learning can create advanced models that are able to generalize with relatively small datasets. This type of task requires a clear architecture to distribute the different types of layers of the AV’s software.
