At automation level 5 as defined by the Society of Automotive Engineers (SAE), a driver will not be in the loop even in a complex driving environment featuring among other challenges, the presence of vehicles with automation levels ranging from 1 (no automation) to 5 (fully automated). This paper defines the safety and ride quality requirements that a fully automated vehicle should meet when operating in a mixed traffic environment featuring vehicles of various automation levels and proposes a Bayesian AI-based driver algorithm as a solution. Design advances that can potentially overcome the safety and ride quality issues are described. Microscopic level data sourced from driving simulator studies are used in applications. Finally, conclusions are presented on the abilities of the Bayesian AI-based driver to meet safety and ride quality criteria while operating in driving environment characterized by uncertainties. The Bayesian AI-based driver is likely to enhance consumer and safety regulator acceptance.

, , , , ,
Advances in Intelligent Systems and Computing
Department of Civil and Environmental Engineering

Khan, A. (2020). Bayesian Artificial Intelligence-Based Driver for Fully Automated Vehicle with Cognitive Capabilities. In Advances in Intelligent Systems and Computing. doi:10.1007/978-3-030-20503-4_6