Accelerating High-Fidelity Road Network Modelling for Indian Traffic Simulations: System Design and Implementation

Abstract

This paper presents the design, architecture, and implementation of IndianRoad.AI, a web-based toolkit developed as part of Smart India Hackathon 2025 Problem Statement SIH25100: Accelerating High-Fidelity Road Network Modeling for Indian Traffic Simulations. The system integrates OpenStreetMap (OSM) data processing, 3D road scene generation, traffic simulation, computer vision-based video analysis, AI-driven smart intelligence, and a multi-stakeholder governance layer into a unified, MATLAB-compatible platform. The toolkit covers six major Indian cities and enables simulation-ready road network analysis with heterogeneous Indian traffic modeling. Experimental results demonstrate the system\'s capability in real OSM file parsing, 3D scene export, mixed-traffic simulation, vehicle detection from live video, intelligent route recommendation, and operational dashboards for traffic police, municipal corporations, smart city administrators, and emergency services.

Introduction

The paper introduces IndianRoad.AI, a web-based platform developed for Smart India Hackathon 2025 (SIH25100) to create high-fidelity, simulation-ready road networks specifically designed for India's complex traffic conditions. Unlike Western traffic systems, Indian roads feature heterogeneous traffic—including cars, motorcycles, auto-rickshaws, buses, and pedestrians—sharing road space with limited lane discipline. Existing traffic simulation tools struggle to accurately model behaviors such as lateral weaving, informal queuing, and intersection encroachment.

Problem Statement

Current road-network modeling approaches face several limitations:

• Existing tools like SUMO assume lane-disciplined traffic and require extensive manual calibration.
• High-definition mapping methods using LiDAR provide accurate results but are too expensive for large-scale deployment.
• No existing open-source platform combines map processing, 3D scene generation, traffic simulation, video analysis, and governance dashboards in a single MATLAB-compatible environment.

IndianRoad.AI addresses these issues by providing an integrated, low-cost, browser-based solution that uses open-source technologies and publicly available data.

System Architecture

The platform is designed as a Single-Page Application (SPA) built with:

• React.js for the frontend
• Chart.js for analytics visualization
• Leaflet.js for interactive maps
• TensorFlow.js with COCO-SSD for computer vision
• OpenStreetMap, GeoJSON, and Overpass API for geospatial data processing

The system consists of four layers:

1. Data Layer – Collects OpenStreetMap data, traffic videos, GPS data, and city datasets.
2. Processing Layer – Performs map parsing, AI-based object detection, traffic simulation, and data fusion.
3. Application Modules Layer – Contains user-facing tools.
4. Output Layer – Exports data as GeoJSON, SUMO XML, OpenDRIVE-compatible files, and MATLAB scripts.

Main Modules

IndianRoad.AI provides ten integrated modules:

• Home
• OSM Analyzer
• 3D Builder
• Traffic Simulator
• Video Analyzer
• Smart Intelligence
• Cities Database
• Analytics
• Authorities Panel
• Emergency Response

All modules share a common data context for maintaining simulation state and workflow continuity.

OSM Analyzer

The OpenStreetMap Analyzer is the primary data-ingestion component. It:

• Imports local OSM files or city datasets.
• Parses road network data using browser-based XML processing.
• Classifies roads into:
  • Highways
  • Arterial Roads
  • Collector Roads
  • Local Roads
• Detects India-specific features such as:
  • Potholes
  • Construction zones
  • Road encroachments
  • Toll plazas

For example, the Bangalore dataset contained:

• 109,563 nodes
• 16,434 road segments (ways)
• Approximately 8,241 km of road network

Export Features

The platform supports exporting simulation-ready data in:

• GeoJSON for GIS applications
• SUMO XML for traffic simulation
• MATLAB scripts for transportation research and modeling

Key Advantages

• Specifically calibrated for Indian traffic behavior.
• Fully browser-based with no server infrastructure required.
• Uses only open-source technologies and public datasets.
• Supports real-time traffic analysis and simulation.
• Provides MATLAB compatibility for academic and research applications.
• Integrates mapping, simulation, AI-based video analysis, and governance tools into a single platform.

Conclusion

This paper presented IndianRoad.AI, a complete implementation of the SIH25100 vision for accelerating high-fidelity road network modeling for Indian traffic simulations. The platform\'s integrated modules collectively automate the pipeline from raw OSM data ingestion and 3D scene generation through heterogeneous traffic simulation, real-time video analysis, AI-driven route intelligence, and multi-stakeholder operational governance dashboards. The implementation demonstrates that a fully browser-based, open-source stack can match the functional capabilities of commercial simulation toolchains for Indian urban contexts at near-zero data acquisition cost. The inclusion of the Authorities Panel — with dedicated views for traffic police, municipal corporations, smart city administrators, and emergency services — positions IndianRoad.AI as a full-stack smart city platform rather than a purely research tool. Future work will integrate satellite imagery-based road segmentation using G-Net architectures [4], expand city coverage to all Tier-1 Indian cities, implement MATLAB Engine API connectivity for bidirectional live data exchange, and incorporate real-time CCTV feed integration for continuous video analysis across multiple city junctions simultaneously.

References

[1] A. Agarwal, A. Kumar, and S. Patel, \"Dynamic traffic signal control for heterogeneous traffic conditions using Max Pressure and Reinforcement Learning,\" Expert Systems with Applications, 2024. [2] F. Xiao et al., \"Advances in reinforcement learning for traffic signal control,\" Intelligent Transportation Infrastructure, 2025. [3] Y. Bao, Y. Cheng, and Z. Liu, \"A review of high-definition map creation methods for autonomous driving,\" Engineering Applications of Artificial Intelligence, 2023. [4] Q. Zhu et al., \"A Global Context-aware and Batch- independent Network for road extraction from VHR satellite imagery,\" ISPRS Journal of Photogrammetry and Remote Sensing, 2021. [5] X. Zhang et al., \"Deep learning-based road extraction from remote sensing imagery: A review,\" ISPRS Journal of Photogrammetry and Remote Sensing, 2025. [6] J. Li et al., \"Maps for Autonomous Driving: Full-process Survey and Benchmark,\" arXiv preprint, 2025. [7] R. Sharma et al., \"Real-time traffic signal optimization for urban mobility,\" Journal of Intelligent Transportation Systems, 2025. [8] M. Singh and K. R. Rao, \"Simulation of signalized intersection with non-lane-based heterogeneous traffic conditions using cellular automata,\" Transportation Research Record, 2024. [9] K. Jeengar et al., \"Multi-objective deep reinforcement learning approach for traffic signal control,\" Accident Analysis and Prevention, 2023. [10] S. Rahmani, M. Sharma, and A. Khan, \"Graph neural networks for intelligent transportation systems: A survey,\" IEEE Transactions on Intelligent Transportation Systems, 2023. [11] D. Krajzewicz, G. Hertkorn, C. Rossel, and P. Wagner, \"SUMO (Simulation of Urban MObility): An open-source traffic simulation,\" Proc. 4th Middle East Symp. Simulation and Modelling (MESM), 2002. [12] A. Dosovitskiy et al., \"CARLA: An open urban driving simulator,\" Proc. 1st Annu. Conf. Robot Learning (CoRL), 2017. [13] TensorFlow.js Team, \"TensorFlow.js: Machine learning for the web and beyond,\" SysML Conference, 2019.

Copyright

Copyright © 2026 Amrutha A Patil, Divya R N, Bhargavi Reddy, Dr. Jyothi Kiran Mirji. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.