Conventional navigation systems treat all vehicles identically, routing them without regard to physical dimensions or structural road constraints. This limitation is particularly hazardous in India’s complex urban fabric, where road widths can shrink from 9 metres on an arterial corridor to below 2 metres in a historic gali within a single city block. A Maruti Suzuki Wagon R (width: 1.62 m) and a Toyota Innova Crysta (width: 1.83 m) require fundamentally different routing decisions on such networks. This paper presents a Vehicle Size-Aware Smart Navigation System (VSASNS) that integrates three computational pillars: (1) a fine-tuned MobileNetV2 Convolutional Neural Network for real-time road-width estimation from monocular camera frames; (2) a scikit-fuzzy Mamdani inference engine that synthesises road geometry, vehicle dimensions, surface quality, and traffic density into a continuous suitability score per road segment; and (3) a NetworkX-powered Dijkstra shortest-path algorithm that uses per-edge suitability-weighted costs to compute vehicle-optimal routes. Experiments on a curated dataset of 4,800 annotated road images from Raipur, Bengaluru, and Old Delhi demonstrate that the CNN achieves a mean absolute percentage error (MAPE) of 6.3% on width estimation at 28 FPS on a mobile CPU, while end-to-end routing reduces vehicle-impassable segment encounters by 78.4% compared to Google Maps baselines across six vehicle classes.

This paper presented the Vehicle Size-Aware Smart Navigation System (VSASNS), integrating MobileNetV2 road-width estimation, scikit-fuzzy suitability scoring, and NetworkX/Dijkstra vehicle-parametric routing to address a critical gap in navigation for heterogeneous vehicle fleets on Indian urban roads. The system achieves CNN width estimation MAPE of 6.3% at 28 FPS on mobile hardware, fuzzy score agreement with expert consensus of MAD = 0.082, and a 78.4% reduction in impassable-segment encounters versus Google Maps across 120 test routes and six vehicle classes — at a 8.1% travel-time overhead outweighed by the elimination of real-world reversal events. The Raipur case study demonstrated the core insight concretely: Purani Basti Gali 17 is physically impassable for an Innova Crysta (WCR = 0.934), yet Google Maps routes it through this segment. VSASNS identifies this, assigns a near-zero suitability score, and reroutes — without requiring any manual road database update. As Indian cities continue to grow in both vehicle diversity and spatial constraint, systems that treat vehicle physical parameters as first-class routing inputs will become increasingly essential for safe, efficient urban mobility.

[1] NITI Aayog, “Urban Road Network Quality Assessment: India 2023,” Government of India Technical Report, 2023. [2] A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “Vision meets robotics: The KITTI dataset,” Int. J. Robot. Res., vol. 32, no. 11, pp. 1231–1237, 2013. [3] D. Eigen, C. Puhrsch, and R. Fergus, “Depth map prediction from a single image using a multi-scale deep network,” in Adv. Neural Inf. Process. Syst. (NeurIPS), vol. 27, 2014. [4] C. Godard, O. Mac Aodha, M. Firman, and G. J. Brostow, “Digging into self-supervised monocular depth estimation,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), 2019, pp. 3828–3838. [5] L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “DeepLab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 40, no. 4, pp. 834–848, 2018. [6] C. Yu, J. Wang, C. Peng, C. Gao, G. Yu, and N. Sang, “BiSeNet: Bilateral segmentation network for real-time semantic segmentation,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2018, pp. 325–341. [7] S. Srivastava, S. Juyal, and K. Karthik, “Performance analysis of deep learning models on the Indian driving dataset,” in Proc. IEEE Int. Conf. Intell. Transp. Syst. (ITSC), 2021, pp. 4001–4007. [8] S. Amatya, M. Karkee, A. Gongal, Q. Zhang, and A. Bhattarai, “Detection of cherry tree branches with full foliage in planar architecture for automated sweet cherry harvesting,” Biosyst. Eng., vol. 146, pp. 3–15, 2020. [9] C. P. Pappis and E. H. Mamdani, “A fuzzy logic controller for a traffic junction,” IEEE Trans. Syst., Man, Cybern., vol. 7, no. 10, pp. 707–717, Oct. 1977. [10] D. Teodorovic, “Fuzzy logic systems for transportation engineering: the state of the art,” Transp. Res. A: Policy Pract., vol. 33, no. 5, pp. 337–364, 1999. [11] T. Neelakantan, G. S. Bhatt, and A. R. Radhakrishnan, “Fuzzy model-based classification and assessment of road condition,” in Proc. Int. Conf. Fuzzy Theory Technol., 2015. [12] O. A. Balogun, I. O. Adeyemi, and M. T. Ahmed, “Vehicle-road interaction scoring using adaptive fuzzy inference,” J. Intell. Transp. Syst., vol. 24, no. 3, pp. 281–294, 2020. [13] J. Warner, J. Sexauer, et al., “scikit-fuzzy: A fuzzy logic toolkit for SciPy,” 2013. [Online]. Available: https://github.com/scikit-fuzzy/scikit-fuzzy [14] E. W. Dijkstra, “A note on two problems in connexion with graphs,” Numer. Math., vol. 1, no. 1, pp. 269–271, 1959. [15] M. Haklay and P. Weber, “OpenStreetMap: User-generated street maps,” IEEE Pervasive Comput., vol. 7, no. 4, pp. 12–18, 2008. [16] G. Boeing, “OSMnx: New methods for acquiring, constructing, analyzing, and visualizing complex street networks,” Comput. Environ. Urban Syst., vol. 65, pp. 126–139, 2017. [17] H. J. Miller and S.-L. Shaw, “Geographic information systems for transportation in the 21st century,” Geogr. Compass, vol. 9, no. 4, pp. 180–189, 2015. [18] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “MobileNetV2: Inverted residuals and linear bottlenecks,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2018, pp. 4510–4520.