Authors : Dr. A. S. Shirkande; Ankita Vitthal Kshirsagar; Aditya Balasaheb Patil; Harshad Nana Shinde

Volume/Issue : Volume 11 - 2026, Issue 10 - October

Google Scholar : https://tinyurl.com/3sr2xzz5

DOI : https://doi.org/10.38124/ijisrt/26Oct684

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : Derailments caused by cracked or fractured rails, shifted track geometry, absent fasteners and intruders on the line remain a serious safety concern, while walking inspections by staff are slow, tiring and hazardous. This review examines the building blocks of a self-driving inspection vehicle controlled by an STM32 microcontroller with embedded artificial intelligence. The vehicle screens the track for eight anomaly types: cracks, fractured rails, obstructions, absent or damaged fittings, geometry misalignment, people or objects on the line, excessive vibration and elevated temperature. Findings are pushed live to a control room, and the vehicle drives itself back to its charging dock once its battery drops below a safe level. We discuss defect categories, sensor choices, neural-network models that fit on Cortex-M parts, energy-aware homing logic and wireless reporting, summarise approaches in tables, and outline unresolved problems and future work.

Keywords : STM32; Railway Inspection; Crack Detection; TinyML; Edge AI; Sensor Fusion; IoT; Autonomous Return; Predictive Maintenance.

References :

1. D. F. Cannon, K.-O. Edel, S. L. Grassie, and K. Sawley, “Rail defects: an overview,” Fatigue Fract. Eng. Mater. Struct., vol. 26, no. 10, pp. 865–886, 2003.
2. R. David et al., “TensorFlow Lite Micro: Embedded machine learning on TinyML systems,” in Proc. MLSys, 2021.
3. C. Banbury et al., “MLPerf Tiny benchmark,” in Proc. NeurIPS Datasets and Benchmarks, 2021.
4. S. Faghih-Roohi, S. Hajizadeh, A. Núñez, R. Babuska, and B. De Schutter, “Deep convolutional neural networks for detection of rail surface defects,” in Proc. IJCNN, 2016, pp. 2584–2589.
5. X. Gibert, V. M. Patel, and R. Chellappa, “Deep multitask learning for railway track inspection,” IEEE Trans. Intell. Transp. Syst., vol. 18, no. 1, pp. 153–164, 2017.
6. J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once: Unified, real-time object detection,” in Proc. IEEE CVPR, 2016, pp. 779–788.
7. S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards real-time object detection with region proposal networks,” in Proc. NIPS, 2015.
8. A. G. Howard et al., “MobileNets: Efficient convolutional neural networks for mobile vision applications,” arXiv:1704.04861, 2017.
9. L. Lai, N. Suda, and V. Chandra, “CMSIS-NN: Efficient neural network kernels for Arm Cortex-M CPUs,” arXiv:1801.06601, 2018.
10. M. G. S. Murshed et al., “Machine learning at the network edge: A survey,” ACM Comput. Surv., vol. 54, no. 8, 2021.
11. B. Jacob et al., “Quantization and training of neural networks for efficient integer-arithmetic-only inference,” in Proc. IEEE CVPR, 2018, pp. 2704–2713.
12. S. Han, H. Mao, and W. J. Dally, “Deep compression: Compressing deep neural networks with pruning, trained quantization and Huffman coding,” in Proc. ICLR, 2016.
13. M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C.
14. Chen,         “MobileNetV2:    Inverted residuals                and         linear bottlenecks,” in Proc. IEEE CVPR, 2018, pp. 4510–4520.
15. J. Redmon and A. Farhadi, “YOLOv3: An incremental improvement,” arXiv:1804.02767, 2018.
16. A. Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv:2004.10934, 2020.