Deep RL Applied to Autonomous Driving

Autonomous driving is one of the hottest technological revolutions in development as of the time of writing this. It will dramatically alter how humanity looks at transportation in general, and will drastically reduce travel costs as well as increase safety. Several state-of-the-art algorithms are used by the autonomous vehicle development community to this end. These include, but are not limited to, perception, localization, path planning, and control. Perception deals with the identification of the environment around an autonomous vehicle—pedestrians, cars, bicycles, and so on. Localization involves the identification of the exact location—or pose to be more precise—of the vehicle in a precomputed map of the environment. Path planning, as the name implies, is the process of planning the path of the autonomous vehicle...