A Survey on Handling Events, Localization and Perception of Autonomous Vehicles in Real Time Scenarios

Abstract

Background and Objective: The ground transportation in the recent decade has seen multiple advancements and yet more advancements are projected with the current pace of introducing innovations. Ideas formulated and introduced in the domain has attempted to revolutionize, improvise, and enhance the applicability of autonomous cars in roadways. Smart vehicles are replacing the normal vehicles, which we are witnessing in various forms of transformations. Cars and trucks are capable of observing the surroundings, take decisions during the runtime and hence adapt to transitioning situations. This feat has been achievable with invention of quick processing sensors and advanced communication technologies. Sensors will be observing the environment for changes, deeming them into relationally impacting situations and based on the perception the autonomous vehicles are advised accordingly. Short range of actions are perceivable and accountable for manual updates by the drivers themselves, and intelligent systems intend to automate the same. Predominant conditions for perceptions are defined based on the measures of safety, traffic, optimization, energy management, ultimately a comfortable drive for the drivers. These measures are achievable metrics when the intelligent systems are concentrating on longer distance perceptions which alters the behavior of vehicles according to the situation. The anticipated problems with such systems are prevalent with even recent advancements and documented to be the noise, manufacturing defects, ambient conditions, and many other factors. The intelligent systems are yet to identify the response to these challenging situations, owing to the unpreparedness of the models to these perceptions. Autonomous vehicles, thus cannot rely upon single sensor and its observations, hence depend on a wide range of sensors for a collective study, incorporated together to provide a 360-degree feedback. This research article contemplates the different techniques available in state-of-art models focused towards the enhancement of autonomous vehicles and intelligent systems. The latest articles are summarized with emphasis on localization of vehicles, perception levels of different models and finally mapping of decisions to AV activities.