A Smartphone Sensor-Based Pavement Roughness Grading Method for Non-Motorized Lanes

DOI: https://doi.org/10.62517/jbdc.202601129

Author(s)

Jiaxin Wei*, Yidan Zhang, Peilun Li, Xinru Guo, Jiahao Han, Jiahao Li

Affiliation(s)

School of Artificial Intelligence and Big Data, Henan University of Technology, Zhengzhou, Henan, China

Abstract

To address the lack of low-cost and routine methods for pavement roughness monitoring on non-motorized lane, this paper proposes a pavement roughness sensing and grading method based on smartphone sensors. Taking shared electric bicycles as the data collection platform, acceleration, gravity, and magnetic field signals are acquired using built-in smartphone sensors, and the vertical acceleration is obtained through coordinate transformation. To reduce low-frequency interference in the raw signals, a zero-phase high-pass filtering and adaptive preprocessing method is applied. The root mean square of vertical acceleration (RMSVA) is used as the main indicator of pavement roughness. In addition, several vibration statistical features are combined to form a multi-dimensional feature set. A support vector machine (SVM) classifier is then used to achieve five-level pavement roughness grading. Experimental results show that the proposed preprocessing method improves the correlation between vibration features and pavement conditions. The SVM classification results show a 93.75% agreement with the RMSVA-based grading results. Field tests on real roads further confirm the stability and effectiveness of the proposed method under different pavement types. The proposed approach is low-cost, easy to deploy, and suitable for large-scale applications. It can provide technical support for the refined maintenance and management of urban non-motorized road networks.

Keywords

Pavement Roughness; Smartphone Sensors; RMSVA; Support Vector Machine; Non-Motorized Lanes

References

[1] Douangphachanh V., Oneyama H. Estimation of road roughness condition from smartphone acceleration measurements. Journal of Traffic and Transportation Engineering, 2015, 2(2):114–123. [2] González A., O’Brien E.J., Li Y.Y., et al. Measurement of pavement roughness using inertial sensors. Automation in Construction, 2008, 17(3):298–305. [3] Jalili F., Ghavami S.M., Afsharnia H. An artificial neural network approach to assess road roughness using smartphone-based crowdsourcing data. Engineering Applications of Artificial Intelligence, 138, 2024:109308. [4] Road pavement health monitoring system using smartphone sensing with a two-stage machine learning model. Automation in Construction, 167, 2024:105664. [5] Abnormal pavement condition detection with vehicle posture data considering speed variations. Sensors, 2024, 24(14):4555. [6] Lee D.Y., Yoo H., Lee J., Jeong G.-O. Smartphone-based assessment of bicycle pavement conditions using road roughness and impact indices. Sustainability, 2025, 17(16):7488. [7] Siriguleng B., Yiliguoqi B., Arong, et al. Research on pavement roughness evaluation method based on smartphones. Journal of China & Foreign Highway, 2025, 45(6):58–66. [8] Guo Shuang, Guo Ying. Adaptive filtering algorithms for multi-channel active noise control. Microprocessors, 2023, 44(5):40–43. [9] Yiliguoqi B., Siriguleng B., Arong, et al. Research on the evaluation and analysis of road surface roughness based on smartphone sensors and SVM. Scientific Reports, 16, 2026:4409. [10]Li Junyi, Yao Yuan, Liu Minghao. SVM-based fault diagnosis of wind turbine gearboxes using optimal vibration feature extraction algorithms. Acta Energiae Solaris Sinica, 2024, 45(7):626–633. [11]He Huitao. Research on damage detection and maintenance strategies for asphalt pavement on expressways. Transport Construction & Management, 2024(6):109–111. [12]Hu Bei. Methods and research on improving asphalt pavement smoothness. Urban Architecture Space, 2022, 29(S1):257–258.