Scholarlink

<p>Potholes present a significant hazard to road safety and infrastructure, making advanced detection systems essential for preventing accidents and minimizing damage. Traditional pothole detection methods often struggle with accuracy, real-time performance, and resource efficiency. This research introduces a novel, low-cost pothole detection approach leveraging smartphone accelerometer and GPS data, enhanced by Dynamic Time Warping (DTW) for robust time series analysis. Data was collected across three diverse road segments-Delhi (2.5&#xa0;km, 917 data points), Srinagar (5&#xa0;km, 1411 data points), and Rajasthan (10&#xa0;km, 2483 data points)-and validated using synchronized video ground truth. The proposed system achieved impressive results, with post-validation accuracy rates of 98.04% in Delhi, 97.02% in Srinagar, and 91.02% in Rajasthan, and a precision rate of 84.05% and recall of 85.71%. Incorporating threshold optimization and additional algorithms for bump and braking detection improved system efficiency by 7.38% and minimized false alarms, ensuring adaptability across varied driving scenarios. This framework not only enhances road safety and maintenance efficiency but also lays the groundwork for broader sensor-based monitoring and integration with smart city infrastructure. Future work will focus on expanding sensor modalities, real-time adaptive learning, and seamless smart city integration.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Smartphone-Sensor Based Dynamic Time Warping Framework for Enhanced Pothole Detection

  • 期刊论文

  • 发表日期：

    2025年8月8日

查看全文
  • Safa Koul,
  • L. Janani,
  • Kartik Asthana,
  • Hyder Mushtaq,
  • Mehvish Alvi,
  • Devashish Kaushik

摘要

Potholes present a significant hazard to road safety and infrastructure, making advanced detection systems essential for preventing accidents and minimizing damage. Traditional pothole detection methods often struggle with accuracy, real-time performance, and resource efficiency. This research introduces a novel, low-cost pothole detection approach leveraging smartphone accelerometer and GPS data, enhanced by Dynamic Time Warping (DTW) for robust time series analysis. Data was collected across three diverse road segments-Delhi (2.5 km, 917 data points), Srinagar (5 km, 1411 data points), and Rajasthan (10 km, 2483 data points)-and validated using synchronized video ground truth. The proposed system achieved impressive results, with post-validation accuracy rates of 98.04% in Delhi, 97.02% in Srinagar, and 91.02% in Rajasthan, and a precision rate of 84.05% and recall of 85.71%. Incorporating threshold optimization and additional algorithms for bump and braking detection improved system efficiency by 7.38% and minimized false alarms, ensuring adaptability across varied driving scenarios. This framework not only enhances road safety and maintenance efficiency but also lays the groundwork for broader sensor-based monitoring and integration with smart city infrastructure. Future work will focus on expanding sensor modalities, real-time adaptive learning, and seamless smart city integration.