AI for Safer Roads: Safe System Speed Assessment Using Explainable AI

Overview

This project was developed for the Asian Development Bank (ADB) AI for Safer Roads Challenge.

The solution evaluates whether existing speed limits are appropriate for the surrounding road environment using:

• Safe System principles
• Explainable AI
• GIS-based risk assessment
• Risk-based intervention prioritization

Key Features

• Road Safety Risk Scoring
• Explainable AI (SHAP)
• Priority Intervention Ranking
• Interactive GIS Risk Map
• Policy-Oriented Recommendations

Repository Structure

• notebooks/ : Main analysis notebook
• outputs/ : Assessment outputs and datasets
• maps/ : Interactive web maps
• visualizations/ : Figures and dashboards
• reports/ : Executive summaries and recommendations

Deliverables

• Interactive Risk Map
• Priority Segments Dataset
• GeoJSON Outputs
• Explainability Analysis
• Policy Recommendations

Methodology

The framework combines:

1. Safe System principles
2. Machine learning-based assessment
3. Explainable AI (SHAP)
4. Risk prioritization
5. Geographic visualization

Challenge Alignment

The objective is not merely to reproduce historical speed limits.

The framework assesses whether current speed limits are appropriate and identifies locations where interventions may improve road safety.

Scalability

The generated spatial outputs (GeoJSON and tabular datasets) are designed to be compatible with enterprise geospatial platforms such as ESRI ArcGIS, supporting future deployment and decision-support workflows.

Documentation

• Findings Summary: docs/ADB_AI4SaferRoads_Findings_Summary.pdf
• Submission Overview: docs/challenge_submission_overview.md
• Main Analysis: notebooks/ADB_AI4SaferRoads_SafeSpeed_Assessment.ipynb

Data Availability

The datasets used in this project were provided under the ADB AI for Safer Roads Challenge Data Use Agreement and are not included in this repository.

See:

data/DATA_NOT_INCLUDED.md