One of the greatest challenges in vehicle dynamics simulations is to accurately represent the tire behavior on different road surfaces. Marco Furlan, Sr. Modelling and Simulation engineer within the Tire Testing and Research group at Calspan recently published work on an innovative method to predict tire friction on different road surfaces.
Using neural networks, this new method relies on tread material properties and the power spectral density of the road surface to predict hysteretic tire friction. Compared to existing approaches, calculations are done in a very computational efficient manner allowing for real-time applications such as Driver-In-the-Loop or Hardware-In-the-Loop simulations.
The paper also demonstrates how data collected at Calspan’s Tire Test Facility, which is world-renowned for its high level of accuracy and repeatability, can be scaled to be representative of virtually any road surface. This opens up an opportunity to simulate vehicle handling performance and motion planning algorithms (for automated driving systems) across a wide range of different road surfaces.
The on-going work and research by Calspan’s Tire Testing and Research group continue to tackle some of the biggest challenges in the development of future mobility. The paper titled “A Neural Network Approach for Roughness-Dependent Update of Tyre Friction” can be found here: https://authors.elsevier.com/a/1eMjS,ZhUEWLlW