Artificial Intelligence Rationale for Autonomous Vehicle Agents Behaviour in Driving Simulation Environment

In recent times, in-vehicle notifications have proliferated with a focus on the exhibition of technological prowess rather than the fulfilment of actual driving needs. In effect, information portrayed by automotive infotainment devices, while useful, is often ignored by the driver due to field of view limitations associated with traditional instrumentation panels. Not surprisingly, under poor visibility conditions and at motorway-level driving speeds, such systems do not effectively present useful information to the user. Contemporary Head-Up Display (HUD) experiments have focused on adapting the aviation-specific characteristics of HUDs to driver-specific needs, obsolescing functionality and simplifying operations where necessary. The more mature approach of these preceding works has revealed that although in-vehicle HUD technological advances have overcome most implementation issues, the related user-centred interface design is in its infancy prohibiting the HUD's unique features from being successfully exploited. Towards addressing this issue, in previous work, we have designed and implemented a functional prototype of a Human Machine Interface (HMI). Specifically, the proposed HMI system introduces a novel design for an automotive HUD interface, which aims to improve the driver’s spatial awareness and response times under low visibility conditions. Particular emphasis has been placed on the prioritisation and effective presentation of information available through vehicular sensors, which would assist, without distracting, the driver in successfully navigating the vehicle. In order to evaluate the effectiveness of the proposed HUD system, we developed a driving simulator based on an open source racing program. Leveraging an open source solution has resulted in manageable levels of incurred expenses whilst allowing us to deliver a flexible simulation application for HUD evaluation. This chapter discusses the artificial intelligence (AI) as developed for the agent vehicles of our open source driving simulator. The simulator was explicitly designed to measure driver’s performance when using the proposed HUD interface and compares its effectiveness to traditional instrumentation techniques. Intuitively, human cognition complexity poses the largest challenge for creating a model of life-like driver’s behaviour for any type of traffic flow. Presuming that specific driving characteristics apply to all human drivers, as dictated by common sense, an attempt was made to form a generic reaction O pe n A cc es s D at ab as e w w w .ite ch on lin e. co m