A High-Tech Solution on the Streets
In early August 2026, the CSIR–Central Road Research Institute (CSIR–CRRI) launched a new indigenous Network Survey Vehicle (NSV) in Delhi. This vehicle is part of a larger project with the Delhi Public Works Department (PWD) to improve urban roads and
reduce dust pollution. Outfitted with an array of modern technology, including high-speed lasers, ultra-high-definition cameras, GPS, and AI-powered analytics, the NSV is designed to conduct rapid and scientific assessments of road conditions. Instead of manual inspections, the vehicle can survey road networks while travelling at normal speeds, gathering data on roughness, surface distress, and other factors without causing major traffic disruptions. The goal is to generate accurate, data-driven insights that help engineers prioritise maintenance and plan for more durable, safer roads across the capital region.
The Challenge of Seeing Potholes
On paper, the technology is a game-changer. Automated systems promise to be more consistent and cost-effective than human inspectors. However, the real world presents numerous challenges for even the most advanced algorithms. A key limitation, noted in specifications for similar systems, is the environment itself. Laser-based sensors, for instance, do not function well on wet surfaces, making post-monsoon surveys difficult. Likewise, video data collection requires good daylight conditions to be effective. The AI models that process this visual data can sometimes be confused by shadows or road debris, misidentifying them as potholes. Some startups in India have had to specifically train their AI to differentiate between a pothole and a pile of garbage to improve accuracy. These systems are constantly learning, but require vast amounts of quality data to become truly reliable under chaotic Indian traffic and environmental conditions.
Visual vs. Vibration
The CRRI vehicle primarily relies on visual and laser-based methods for detection. This approach is excellent for cataloguing visible cracks, road markings, and geometry. However, other emerging technologies use a different method: vibration. Some companies use smartphone apps in vehicles to record vibrations, flagging any spot where a jolt exceeds a certain threshold as a potential problem. The advantage of this approach is that it can work even when a pothole is filled with water or otherwise obscured from a camera's line of sight. While the NSV provides a comprehensive surface scan, capturing everything from pavement texture to roadside assets, the detection of every single pothole remains a complex task that no single technology has perfected. Even with high-definition cameras, a layer of human verification is often needed to confirm the AI's findings.
Beyond Detection: The Repair Pipeline
Ultimately, detecting a pothole is only the first step. The true measure of success is how quickly and effectively it gets repaired. The data generated by the NSV is intended to feed into a Road Asset Management System (RAMS) to help authorities prioritize work. Yet, this is where technology meets bureaucracy. The process involves multiple agencies, budget allocations, and contractor timelines. Even with perfect, real-time data on every road defect, the institutional capacity to act on that information must also be robust. No amount of AI can fix a pothole if there is a lag in the maintenance pipeline. The NSV is a powerful tool for identifying problems scientifically, but it cannot, by itself, overhaul the complex logistics of road repair in a megacity like Delhi.
