Toward a Cooperative ADAS for Train Driving based on Real-Time Human Parameters and Delay Estimation

Abstract

Modern railway research focuses on autonomous train control systems, however the existing rolling stock still requires human driver in the control loop, especially for safety critical operations. This article is about the interaction of the train driver and an autonomous driver advisory system (ADAS) based on human state estimation to avoid confrontative and encourage integrative cooperation modes. This is made possible using a special observer design approach that allows to estimate driver's reaction delay to ADAS advises, and to reconfigure the ADAS mode accordingly. The result is that the driver acceptance rates (for ADAS suggestions) improve, along with the overall train control performance, ensuring successful mission. Theoretical developments and simulation results are provided.