FixMyPath: Smartphone Sensor-Based Pothole Detection Using Machine Learning

This paper presented FixMyPath, a smartphone- accelerometer-based road-surface monitoring system that integrates a trained Random Forest classifier with a fully functional Android application. Vibration data collected across three surface categories—normal road, speed breaker, and pothole—were processed into orientation-invariant magnitude features, enabling robust classification irrespective of handset mounting position. The classifier achieved an overall accuracy of 94.3 %, with per-class F1- scores confirming reliable discrimination across all categories. The companion application performs inference at a mean latency of 18 ms and sustains low battery overhead, making continuous passive monitoring feasible during everyday commutes. The proposed framework offers a cost-effective, scalable complement to conventional road-inspection workflows. By transforming the ubiquitous smartphone into a pothole sensor, FixMyPath enables large-scale, citizen-driven data collection that could supply municipal agencies with timely, high-resolution road-condition intelligence at negligible cost. Future work will pursue four directions. First, GPS integration will be incorporated to attach geographic coordinates to each detection event, enabling spatial aggregation and the construction of city-scale pothole heat maps. Second, the training dataset will be expanded to encompass a wider variety of vehicle categories— motorcycles, auto-rickshaws, and cars—and additional pavement types prevalent across India. Third, deep learning architectures, particularly Long Short-Term Memory (LSTM) networks, will be evaluated as potential replacements for the Random Forest model to leverage temporal dependencies in the vibration signal. Fourth, a cloud-based dashboard will be developed to aggregate crowdsourced reports, apply spatial clustering algorithms to filter noise, and present decision-ready maintenance recommendations to road authorities.

[1] P. Mohan, V. N. Padmanabhan, and R. Ramjee, “Nericell: Rich monitoring of road and traffic conditions using mobile smartphones,” in Proc. ACM SenSys, Raleigh, NC, USA, 2008, pp. 323–336. [2]   J. R. Kwapisz, G. M. Weiss, and S. A. Moore, “Activity recognition using cell phone accelerometers,” ACM SIGKDD Explor. Newsl., vol. 12, no. 2, pp. 74–82, 2011. [3]   A. Mednis, G. Strazdins, R. Zviedris, G. Kanonirs, and L. Selavo, “Real-time pothole detection using Android smartphones with accelerometers,” in Proc. IEEE DCOSS, Barcelona, Spain, 2011, pp. 1–6. [4]  L. Breiman, “Random forests,” Mach. Learn., vol. 45, no. 1, pp. 5–32, 2001. [5]  R. Sharma, A. Patel, and S. Verma, “Smartphone-based road surface monitoring using machine learning,” Int. J. Intell. Transp. Syst. Res., vol. 22, no. 1, pp. 45–58, 2024. [6]  S. Eriksson, H. Sjögren, and J. Lindqvist, “Pothole detection system: A machine learning approach,” Int. J. Smart Sens. Intell. Syst., vol. 11, no. 1, pp. 1–18, 2018. [7] F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011. [8]  Y. Asakura and M. Kashiwadani, “Road network reliability caused by daily fluctuation of traffic flow,” in Proc. 19th PTRC Summer Annual Meeting, Brighton, UK, 1991, pp. 73– 84. [9]  K. de Zoysa, C. Keppitiyagama, G. P. Seneviratne, and W. W. A.   T. Shihan, “A public transport system based sensor network for road surface condition monitoring,” in Proc. ACM SIGCOMM Workshop on Networked Systems for Developing Regions, Kyoto, Japan, 2007, pp. 1–6. [10]  Ministry of Road Transport and Highways, Government of India, “Annual Report on Road Accidents in India,” New Delhi, India, 2022. [Online]. Available: https://morth.nic.in