In traction control and active safety, accurate vehicle models are essential to predict vehicle dynamics and act accordingly. Some of the most important elements of these models cannot be measured directly, and must be estimated based on measurable data.

• One FlexCase is being used to collect vehicle dynamics data and to run a machine learning algorithm to estimate the lateral and longitudinal components of velocity. This project went from Simulink model to real-world testing in a few hours.
• Another FlexCase measures dynamics and powertrain data to estimate the frictional properties of the tires. The estimation algorithm is trained on data produced by smart sensors, also developed at the lab.

The FlexCase’s portability allows the ML algorithms to be quickly integrated into different vehicle platforms for more training data and validation.