Model-based observers for vehicle dynamics and tyre force prediction

Advanced control and driving assistance systems play a major role in modern vehicles, ensuring higher standards of safety and performance. Their correct operation extensively depends on the knowledge of tyre forces and vehicle drift. However, these quantities are hard to measure directly, due to cost or technological reasons. One possible alternative that is attracting much attention in the last few years is represented by virtual sensing where the quantities of interest can be inferred using a physical model that maps the relationship between these quantities and other available direct measurements, like accelerations, velocities and rate-of-turns. In this research, model-based observation is adopted to predict tyre forces and slip angles. In contrast to existing systems, ours relies on direct causality equations without the need of any explicit tyre model. Different observers are developed that are grounded, respectively, in the Cubature Kalman and Particle filtering, and they are contrasted against the standard Extended Kalman filter (EKF). Results are presented to quantitatively assess the performance of the observers using a 14 Degrees Of Freedom (DOFs) full vehicle model that has been simulated in standard manoeuvres including constant radius cornering, increasing and swept-sine steering, and sine-dwell manoeuvring. Although all three embodiments allow model nonlinearities and measurement noise to be appropriately tackled, the two Kalman filters outperform the PF in terms of estimation accuracy, especially for tyre force prediction. In addition, the novel Cubature Kalman filter shows comparable accuracy and robustness, but higher stability when compared to the EKF.

Automated summary

Advanced control and driving assistance systems in modern vehicles ensure higher safety and performance standards. Virtual sensing using physical models to infer tyre forces and slip angles is a promising alternative. Model-based observation, particularly with Kalman filters, shows better accuracy and stability compared to traditional methods like Extended Kalman filter.