Model based vertical dynamics estimation with Modelica and FMI

This paper analyses the performance of Modelica implemented state estimation algorithms for semi-active suspension control for the DLR ROboMObil (ROMO). In this approach the prediction model for the vertical dynamics state estimation and the tire contact force estimation is designed as a quarter vehicle model, which directly incorporates all relevant nonlinear parts. Based on this prediction model a square root unscented Kalman-filter (SR-UKF) is implemented, using the DLR Modelica Kalman-filter library and the Functional Mockup Interface (FMI). In a consecutive step this prediction model is extended by introducing an input port for road obstacle information, e. g. extracted from image data from ROMO 360 degree stereo surround view. The observer design and implementation on real-time hardware are performed in Modelica using the automated tool chain from the Modelica simulator to the Rapid Control Prototyping (RCP) hardware. Experimental results from a four post test-rig and simulations illustrate, that the estimation accuracy can be improved by the SR-UKF compared to an extended Kalman filter (EKF) based implementation.