Background As the automotive industry progresses toward Level 5 Autonomous Vehicles, traditional mechanical steering systems are being replaced by electronic and sensor-driven steering mechanisms. The precision and performance of steering controllers have become increasingly critical, as minor input oscillations can lead to perceptible noise and vibrations. Such disruptions impact both vehicle comfort and safety, underscoring the need for innovative solutions. Method This study evaluates the noise and vibration behavior in autonomous vehicle steering systems. Specifically, it investigates the role of bearings during full steering rotations ranging from −450° to 450°, performed at steering rates of 50–100°/s. A novel end-of-line (EOL) acoustic analysis technique was introduced to detect noise-related failures through frequency domain analysis. The methodology employs high- and low-frequency filters to isolate and identify failure modes associated with:<UnorderedList Mark="Bullet"> <ItemContent> Steering motor harmonics </ItemContent> <ItemContent> Bearing wear caused by press-fit-induced defects </ItemContent> </UnorderedList> Result The analysis concluded that:<UnorderedList Mark="Bullet"> <ItemContent> High-frequency noise is primarily generated by steering motors. </ItemContent> <ItemContent> Low-frequency noise (100–200 Hz) predominantly originates from defects in the lower bearing, specifically due to press-fit-induced wear. </ItemContent> </UnorderedList> By employing the newly developed acoustic analysis method, defect detection efficiency improved by up to 90%. Additionally, the diagnostic technique demonstrated robustness and independence from system configurations, achieving a validated confidence level of 90% across diverse steering setups and bearing types. Conclusion This study introduces a scalable and reliable acoustic method for noise-based failure detection in autonomous vehicle steering systems. By enhancing defect identification during the production process, the approach supports quality assurance and enables early fault detection. Consequently, the method contributes to the safe and seamless operation of advanced steering architectures within fully autonomous vehicles.

Autonomous Vehicle Steering Noise Diagnosis Using a Novel End-of-Line Acoustic Screening Methodology

期刊论文
发表日期：

2025年8月15日

查看全文

Mahesh Kumar Balthy,
Roshan Raman,
Rohit Singh Lather

摘要

Background

As the automotive industry progresses toward Level 5 Autonomous Vehicles, traditional mechanical steering systems are being replaced by electronic and sensor-driven steering mechanisms. The precision and performance of steering controllers have become increasingly critical, as minor input oscillations can lead to perceptible noise and vibrations. Such disruptions impact both vehicle comfort and safety, underscoring the need for innovative solutions.

Method

This study evaluates the noise and vibration behavior in autonomous vehicle steering systems. Specifically, it investigates the role of bearings during full steering rotations ranging from −450° to 450°, performed at steering rates of 50–100°/s. A novel end-of-line (EOL) acoustic analysis technique was introduced to detect noise-related failures through frequency domain analysis. The methodology employs high- and low-frequency filters to isolate and identify failure modes associated with:

Steering motor harmonics

Bearing wear caused by press-fit-induced defects

Result

The analysis concluded that:

High-frequency noise is primarily generated by steering motors.

Low-frequency noise (100–200 Hz) predominantly originates from defects in the lower bearing, specifically due to press-fit-induced wear.

By employing the newly developed acoustic analysis method, defect detection efficiency improved by up to 90%. Additionally, the diagnostic technique demonstrated robustness and independence from system configurations, achieving a validated confidence level of 90% across diverse steering setups and bearing types.

Conclusion

This study introduces a scalable and reliable acoustic method for noise-based failure detection in autonomous vehicle steering systems. By enhancing defect identification during the production process, the approach supports quality assurance and enables early fault detection. Consequently, the method contributes to the safe and seamless operation of advanced steering architectures within fully autonomous vehicles.