Longitudinal Vehicle Speed Estimation for Four-Wheel-Independently-Actuated Electric Vehicles Based on Multi-Sensor Fusion

Xiaolin Ding,Zhenpo Wang,Lei Zhang,Cong Wang

DOI: https://doi.org/10.1109/tvt.2020.3026106

IF: 6.8

2020-11-01

IEEE Transactions on Vehicular Technology

Abstract:In this paper, an enabling multi-sensor fusion-based longitudinal vehicle speed estimator is proposed for four-wheel-independently-actuated electric vehicles using a Global Positioning System and Beidou Navigation Positioning (GPS-BD) module, and a low-cost Inertial Measurement Unit (IMU). For accurate vehicle speed estimation, an approach combing the wheel speed and the GPS-BD information is firstly put forward to compensate for the impact of road gradient on the output horizontal velocity of the GPS-BD module, and the longitudinal acceleration of the IMU. Then, a multi-sensor fusion-based longitudinal vehicle speed estimator is synthesized by employing three virtual sensors which generate three longitudinal vehicle speed tracks based on multiple sensor signals. Finally, the accuracy and reliability of the proposed longitudinal vehicle speed estimator are examined under a diverse range of driving conditions through hardware-in-the-loop tests. The results show that the proposed method has high estimation accuracy, robustness, and real-time performance.

telecommunications,engineering, electrical & electronic,transportation science & technology

What problem does this paper attempt to address?

This paper attempts to solve the problem of inaccurate longitudinal vehicle speed estimation of four - wheel - independently - actuated electric vehicles (FWIA EVs) under different driving conditions. Specifically, the paper proposes a longitudinal vehicle speed estimation method based on multi - sensor fusion, aiming to improve the estimation accuracy and robustness, especially when using low - cost sensors.

### Main problems
1. **Longitudinal vehicle speed cannot be directly measured**: Traditional low - cost sensors cannot directly measure the longitudinal vehicle speed. Especially in four - wheel - independently - actuated electric vehicles, there is no driving wheel from which the vehicle speed can be directly obtained by wheel speed integration.
2. **Sensor errors and noise**: Low - cost sensors such as IMU and GPS - BD modules have bias and noise problems, which affect the accuracy of vehicle speed estimation.
3. **Influence of road grade**: Road grade will have an impact on the horizontal speed output by the GPS - BD module and the longitudinal acceleration measured by the IMU, and compensation is required.
4. **Reliability under complex driving conditions**: How to maintain high - precision and real - time vehicle speed estimation under different driving conditions (such as sudden acceleration, sudden braking, low - adhesion road surfaces, etc.).

### Solutions
To solve the above problems, the paper proposes the following methods:
1. **Multi - sensor fusion**: Combine multi - source information such as GPS - BD module, IMU, wheel speed sensor, etc., and improve the accuracy and robustness of vehicle speed estimation through multi - sensor fusion algorithms.
2. **Road grade estimation**: By combining wheel speed and GPS - BD measurement results, a road grade estimation method is proposed to calibrate the IMU bias and improve the accuracy of GPS - BD speed.
3. **Virtual sensor design**: Design three virtual sensors. Each virtual sensor generates a longitudinal vehicle speed trajectory, and then filters out the noise through the Kalman filter.
4. **Central fusion filter**: Use the central fusion filter to fuse the three vehicle speed trajectories, adjust the weight coefficients according to the covariance matrices of each trajectory, and finally obtain an accurate vehicle speed estimate.

### Verification and evaluation
To verify the effectiveness of the proposed method, the paper carried out hardware - in - the - loop (HIL) tests, covering a variety of typical driving conditions, including:
- Road grade estimation
- GPS - BD signal loss (high - adhesion road surface)
- Sudden acceleration and sudden braking on low - adhesion road surfaces
- Sudden acceleration and sudden braking on low - adhesion road surfaces and GPS - BD signal loss
- Sudden acceleration on low - adhesion ramps

The experimental results show that the proposed multi - sensor - fusion longitudinal vehicle speed estimation method has high estimation accuracy, computational efficiency and reliability, and performs well under various driving conditions.

### Formula summary
1. **Longitudinal acceleration measured by IMU**:
\[
a_{\text{IMU}}=\dot{v}_x-\omega_r\cdot\dot{v}_y + g\cdot\sin(\theta)+\sigma+\tau
\]
where $\sigma$ is the IMU bias and $\tau$ is the random measurement noise.

2. **Road grade estimation**:
\[
\theta=\arcsin\left(\frac{a_{\text{filt}}-a_{\text{wheel}}}{g}\right)
\]

3. **Corrected longitudinal acceleration**:
\[
a_r = a_{\text{filt}}-g\cdot\sin(\theta)
\]

4. **Speed trajectory of virtual sensor 1**:
\[
\text{Track\_v1}=v_{\text{Tk}}^{\text{GPS}}+N\sum_{i = 0}^{N}(a_{r,k}\cdot\zeta)
\]

5. **Speed trajectory of virtual sensor 2**:
\[
\text{Track\_v2}=\frac{\pi\cdot n_w}{30}\cdot R_e
\]

6. **Speed trajectory of virtual sensor 3**:
\[
\text{Track\_v3}=\hat{}