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

摘要

Advances in Robotics, Automation and Control 316 pattern from a list of possible reactions to given situations. The AI embedded in participating vehicles in the simulations embraced certain assumptions, which are outlined later in the chapter. In this way the development time required for the AI routines was reduced by discounting rare occurrences that would indicate a severely misbehaving and irrational human driver. However, it was deemed necessary to develop AI controlled vehicles that could perform potential human miscalculations in order to enhance the realism of the simulation scenarios and their degree of mapping to real-life situations,. Imitation of human error by the robot vehicles, such as failure to brake on time, became a simulation feature that substantially enhanced the driver's immersion in the synthetic environment. Overall, this chapter elaborates on the modelling process of the agent vehicle's AI and discusses contemporary attempts to develop similar AI simulations for other simulation facilities. Notably, throughout this work, a concerted effort was made to ensure that the simulator scenarios were fair representations of potential real-world accident-prone situations; this has been made possible after attentive examination of the statistical data on driving scenarios provided by the Strathclyde Police Department situated in the City of Glasgow, UK.