Editorial Type:
Article Category: Research Article
 | 
Online Publication Date: 09 Nov 2023

Improved Vehicle Longitudinal Velocity Estimation Using Accelerometer Based Intelligent Tire

,
,
,
, and
Page Range: 182 – 209
DOI: 10.2346/tire.22.20012
Save
Download PDF

ABSTRACT

An intelligent tire–based algorithm was developed to reinforce the vehicle longitudinal velocity estimation, from the vehicle inertial measurement unit (IMU). A tire was instrumented using a triaxis accelerometer (intelligent tire) in an instrumented vehicle with an IMU, and a global positioning system (GPS) based speed sensor (VBOX) as the ground truth for vehicle velocity. A testing matrix was developed, including two tire inflation pressures, two normal loads, and variable speed between 4 m/s to 14 m/s. A signal processing algorithm was developed to analyze the data from the accelerometer. Variational mode decomposition and Hilbert spectrum analysis were used for extracting features from each tire revolution. Later, a machine learning algorithm was trained to estimate the velocity using the acceleration data from the intelligent tire. Because the sampling rates of the IMU data and the intelligent tire data were different, sensor fusion was implemented. This calculated velocity was then used to correct the IMU-based estimated velocity. This new velocity can be used to enhance the performance of all advanced chassis control systems, such as anti-lock braking system (ABS) and electronic stability program (ESP).

Keywords: intelligent tire; machine learning; signal processing; regression; velocity estimation; sensor fusion
FIG. 1 —
FIG. 1 —

Triaxis accelerometer installed in the central lining of the tire.

FIG. 2 —
FIG. 2 —

Instrumented 2003 VW Jetta with an intelligent tire.

FIG. 3 —
FIG. 3 —

x, y, and z axis intelligent tire raw data.

FIG. 4 —
FIG. 4 —

x, y, and z axis intelligent tire data for one revolution.

FIG. 5 —
FIG. 5 —

z axis intelligent tire data for different velocities.

FIG. 6 —
FIG. 6 —

Effect of inflation pressure of tire on contact patch length.

FIG. 7 —
FIG. 7 —

Effect of loading conditions on contact patch length.

FIG. 8 —
FIG. 8 —

Effect of loading conditions on z axis intelligent tire signal power.

FIG. 9 —
FIG. 9 —

Effect of inflation pressure of tire on z axis intelligent tire signal power.

FIG. 10 —
FIG. 10 —

VMD on z axis data of intelligent tire.

FIG. 11 —
FIG. 11 —

Last mode of VMD on x axis data of intelligent tire.

FIG. 12 —
FIG. 12 —

Last mode of VMD on z axis data of intelligent tire.

FIG. 13 —
FIG. 13 —

HHT on z axis data of intelligent tire.

FIG. 14 —
FIG. 14 —

Last mode of VMD seen in HHT on z axis data of intelligent tire.

FIG. 15 —
FIG. 15 —

Last mode of VMD seen in HHT on z axis data of intelligent tire for 13.9 m/s (50 kph) velocity.

FIG. 16 —
FIG. 16 —

Last mode of VMD seen in HHT on z axis data of intelligent tire for 8.3 m/s (30 kph) velocity.

FIG. 17 —
FIG. 17 —

Last mode of VMD seen in HHT on z axis data of intelligent tire for 4.1 m/s (15 kph) velocity.

FIG. 18 —
FIG. 18 —

Feature extraction process.

FIG. 19 —
FIG. 19 —

Scatter plot of velocity estimated by machine learning model.

FIG. 20 —
FIG. 20 —

Predicted velocity versus true velocity scatter plot.

FIG. 21 —
FIG. 21 —

IMU-based velocity estimation.

FIG. 22 —
FIG. 22 —

IMU-based velocity estimation error.

FIG. 23 —
FIG. 23 —

IMU-based and intelligent tire–based velocity estimation.

FIG. 24 —
FIG. 24 —

Sensor fusion of IMU and intelligent tire data using Kalman filter.

FIG. 25 —
FIG. 25 —

IMU and intelligent tire fused velocity estimation.

FIG. 26 —
FIG. 26 —

IMU and intelligent tire fused velocity estimation error.

FIG. 27 —
FIG. 27 —

IMU and intelligent tire fused velocity estimation error (before IMU-based velocity drift).

FIG. 28 —
FIG. 28 —

Boxplot of IMU and intelligent tire fused velocity estimation error.

FIG. 29 —
FIG. 29 —

Boxplot of IMU and intelligent tire fused velocity estimation error (before IMU-based velocity drift).

Contributor Notes

1 Department of Mechanical Engineering, Virginia Tech, Randolph Hall, 460 Old Turner Street, Blacksburg, Virginia 24060, USA
2 Corresponding author. Email: rajvardhan@vt.edu
  • Download PDF