ADB Report on Predicting Road Quality and Accessibility: Reimagining Infrastructure Monitoring with AI and Big Data
SDG 9: Industry, Innovation, and Infrastructure | SDG 10: Reduced Inequalities | SDG 11: Sustainable Cities and Communities
Ministry of Road Transport and Highways (MoRTH) | NHAI | NITI Aayog | National Statistical Office (NSO)
The ADB Report on Predicting Road Quality and Accessibility (December 2025) highlights a critical shift from traditional, resource-intensive road monitoring to AI-driven and geospatial assessment models. While roads are recognized as development lifelines, chronic underfunding in maintenance—with a funding gap of up to $1.9 billion in some developing economies—often leads to structural failures that cost three times more than regular upkeep.
Key technical and strategic innovations detailed in the report include:
The Rural Accessibility Index (RAI): A vital metric for SDG Indicator 9.1.1, measuring the share of the rural population living within 2 km of an all-season road.
Machine Learning Breakthroughs: Transfer learning and Convolutional Neural Networks (CNNs) are achieving up to 94% accuracy in predicting road quality from satellite imagery, offering a scalable alternative to physical surveys.
Smartphone-Based Monitoring: Leveraging the 3-axis accelerometers in modern smartphones to detect road roughness and potholes through vehicle vibration, significantly lowering data collection costs.
Hybrid Assessment Systems: A proposed framework where satellite imagery is used for initial broad-category screening (good, satisfactory, poor), followed by detailed roughness assessments via smartphones or laser profilers for prioritized segments.
Data Revolution Barriers: The transition faces challenges including GPS limitations in remote areas, data privacy risks with crowdsourced location data, and the high cost of purchasing high-resolution satellite imagery for granular assessments.
What is the International Roughness Index (IRI)? It is a globally standardized metric used to measure pavement roughness by simulating the suspension movement of a “quarter-car” vehicle model over a road surface. Expressed in meters per kilometer (m/km), a lower IRI indicates a smoother road suitable for high speeds, while a high IRI signals significant distress, such as potholes and rutting, that increases vehicle maintenance costs and fuel consumption.
Policy Relevance
The ADB’s findings are highly significant for India’s Gati Shakti National Master Plan and the PMGSY (Pradhan Mantri Gram Sadak Yojana). With India’s Rural Accessibility Index (RAI) estimated at 75.2%, nearly a quarter of the rural population still lacks all-season connectivity, directly impacting poverty reduction and access to healthcare. Adopting a “Smartphone-as-a-Sensor” model could democratize road quality monitoring across India’s 6.3 million km road network, enabling the NSO to move SDG 9.1.1 reporting from a Tier 2 to a Tier 1 status. Furthermore, integrating AI-driven predictive maintenance into the NHAI’s Data Lake would allow for “preventive” rather than “reactive” repairs, potentially saving thousands of crores in long-term reconstruction costs while enhancing disaster resilience in climate-vulnerable regions like the Himalayas.
Follow the full report here: PREDICTING ROAD QUALITY AND ACCESSIBILITY INDICATORS THROUGH CONVENTIONAL AND INNOVATIVE METHODS