Test-Driven Full Vehicle Modelling for ADAS Algorithm Development

ADAS system development for safety as well as comfort is a major activity for all AOEMs and dedicated system suppliers. The development of these systems relies heavily on the availability of accurate driving dynamics models to validate the control algorithms and verify vehicle performance in realistic driving scenarios. AOEMs usually have access to high-fidelity full vehicle models but the availability of such models poses a significant challenge to software and sub-system suppliers. In order for them to develop their ADAS solutions, an accurate test-based model identification process using prototype or benchmark vehicles would be highly desirable. In this paper, a stepwise approach for the identification of the vehicle parameters in order to create an accurate 15-DOF vehicle dynamics model is proposed. The approach relies on an optimized use of vehicle driving test data to minimize test bench or laboratory testing. The test data is used to identify the various vehicle parameters such as inertia parameters, suspension stiffness and damping and tire stiffness. Subsequently the model is validated through simulation and test. Finally, the use of the model in context of ADAS development is illustrated by integrating it in a traffic scenario simulator allowing MiL/SiL/HiL controls validation for various driving scenarios such as a lane change or autonomous valet parking.