Autonomous vehicles are most concerned about safety control issues, and the slip ratio is critical to the safety of the vehicle control system. In this paper, different machine learning algorithms (Neural Networks, Gradient Boosting Machine, Random Forest, and Support Vector Machine) are used to train the slip ratio estimation model based on the acceleration signals ($a_x$, $a_y$, and $a_z$) from the tri-axial Micro-Electro Mechanical System (MEMS) accelerometer utilized in the intelligent tire system, where the acceleration signals are divided into four sets ($a_x/a_y/a_z$, $a_x/a_z$, $a_y/a_z$, and $a_z$) as algorithm inputs. The experimental data used in this study are collected through the MTS Flat-Trac tire test platform. Performance of different slip ratio estimation models is compared using the NRMS errors in 10-fold cross-validation (CV). The results indicate that NN and GBM have more promising accuracy, and the $a_z$ input type has a better performance compared to other input types, with the best result being the estimation model of the NN algorithm with $a_z$ as input, which results is 4.88%. The present study with the fusion of intelligent tire system and machine learning paves the way for the accurate estimation of tire slip ratio under different driving conditions, which will open up a new way of Autonomous vehicles, intelligent tires, and tire slip ratio estimation.
Download