From Reactive to Predictive: India's Bet on Data-Driven Road Safety

News

1. Anthropic Poaches Microsoft AI Executive for Infrastructure Role

2. Uber Taps AWS Chips for Ride Matching, AI Training

3. OpenAI Urges Grid Upgrades, Stronger Safety Nets for AI Age

4. Wipro Buys Mindsprint in $1 Billion-Plus Olam Deal

5. Magicpin Fast-Tracks AI Rollout as LPG Crisis Disrupts Restaurants

6. Coforge Names Sunil Fernandes COO Amid AI-Native Push

India’s road safety crisis is worsening as traffic grows heavier and more unforgiving. The latest available government data show road accidents rose from 372,181 in 2020 to 487,705 in 2024, while fatalities climbed from 138,383 to 177,177 over the same period.

Behind those numbers is a pattern that has proved stubbornly resistant to slogans and sporadic crackdowns. 

Overspeeding remained the biggest single factor in 2024, linked to 344,446 accidents and 123,947 deaths. Basic safety lapses were just as deadly. Deaths linked to not wearing helmets reached 54,493 last year, while seat belt-related fatalities stood at 14,595.

Accidents involving people without a valid driving license also remained a significant problem.

A new approach is emerging that seeks to tackle the crisis through data-driven governance, analytics and artificial intelligence rather than isolated interventions.

Building the Data Backbone 

At the center of this shift is the Ministry of Road Transport and Highways’ push to build a national digital architecture for crash reporting, analysis and intervention through eDAR, or the Electronic Detailed Accident Report, which sits alongside the earlier iRAD framework.

The platform is designed to bring police, transport, health and insurance inputs into a single workflow that supports both prevention efforts and claims processing. The eDAR portal lays out roles for police, hospitals, transport authorities, insurers, district officials and highway agencies, reflecting the breadth of the reporting chain.

The ambition is straightforward but difficult to execute: turn scattered accident records into a system that shows where crashes are happening, why they recur and which interventions are most likely to work.

That matters in a country where a single crash can reflect a mix of driver behavior, vehicle condition, road design, enforcement gaps and emergency response time.

Once aggregated, the data can help authorities identify black spots, behavioral trends and infrastructure failures that might otherwise remain buried in local records, allowing interventions to be aimed more precisely through engineering changes, targeted enforcement or faster trauma response.

Data-Driven Governance

For decades, many road-safety interventions in India were guided by local administrative judgment, complaints or after-the-fact reactions. Access to digital dashboards is beginning to change that.

District authorities can now analyze patterns across crashes and respond more systematically, using data to guide decisions on road design, enforcement and post-crash care. 

Policymakers do not always frame these tools as artificial intelligence in the popular sense. Often, the language is more restrained: analytics, modeling and decision-support systems.

“The intelligence lies in how the data is analyzed,” said Pankaj Mehra, a road safety researcher working with state governments on analytics-led intervention. 

The larger shift is from person-dependent decision-making to process-led intervention, with action guided less by instinct and more by structured evidence.

Odisha’s Experiment

One of the clearest examples is emerging in Odisha, where researchers from IIT Madras have partnered with the state government on a data-driven road safety strategy.

The collaboration was formalized through a memorandum of understanding signed on 23 December 2024 between IIT Madras’ RBG Labs and Odisha’s Commerce and Transport Department.

IIT Madras said the effort is guided by its 5E framework of Engineering, Education, Enforcement, Emergency Care, and Empathy.

The aim is to use data and evidence-based policy to reduce accidents and fatalities while providing the state with a more structured way to identify and respond to risks.

“With too much data and too little time, AI helps convert complex road safety information into meaningful insights,” Mehra said.

Despite the analytical machinery behind it, the program’s public-facing message is simple: Come Home Safe.

Driver Education and AI

Technology and infrastructure can reduce risk, but road safety still turns heavily on human behavior. That is where IIT Madras is trying a different approach.

This February, the institute’s Center of Excellence for Road Safety introduced ThinnAI, an AI-enabled personalized trainer aimed at improving driver readiness before new drivers enter the licensing pipeline. IIT Madras said the platform was unveiled during the India AI Impact Summit 2026 in Delhi and is designed to move beyond rote memorization toward practical preparedness.

The platform uses adaptive assessments and interactive modules to evaluate drivers’ traffic knowledge, risk perception, and situational judgment. The idea is not simply to help applicants clear a test, but to build safer driving habits earlier.

“Behavioral discipline forms with habit. Driving must become a responsible habit,” Venkatesh Balasubramanian, Head of Centre of Excellence in Road Safety at IIT Madras, said.

He opined that safe driving demands higher-order cognitive abilities such as risk perception, situational awareness and sound judgment.

The name draws on the traditional thinnai, a front-courtyard space in many South Indian homes where elders once passed on social knowledge and norms. The digital version is meant to do something similar for road behavior, turning driver education into a process of gradual learning rather than a one-time exam.

Mehra said effective training works best in three stages: understanding the rules, learning vehicle control and then gaining real-world driving experience. Most licensing systems emphasize the second stage while neglecting the first. ThinnAI aims to strengthen that missing layer.

Emerging Technologies

The legal framework for more technology-led enforcement has been in place since the Motor Vehicles (Amendment) Act, 2019, which opened the way for electronic monitoring and enforcement.

At the India AI Impact Summit in February, road-safety officials and researchers pointed to automated systems that can detect violations through cameras, generate digital evidence and analyze accident patterns with greater consistency. 

Another area drawing attention is vehicle-to-vehicle communication, where cars exchange real-time warnings about hazards, sudden braking or possible collisions.

Taken together, such systems point to a road-safety model that is more connected, more preventive and less reliant on fragmented local reporting.

The Two-Wheeler Challenge

Any Indian road-safety strategy also has to confront a basic fact of the country’s traffic mix: two-wheelers account for a disproportionate share of deaths. The latest government data show they were involved in 45% of road fatalities in 2024, while pedestrians accounted for about 20%.

That makes India different from many markets where road-safety innovation is designed mainly around cars. Here, the challenge is broader, spanning helmets, rider behavior, visibility, road design and emergency response.

Recognizing that enforcement and technology alone will not be enough, the government has also rolled out the MITRA Road Safety Program with the Union Ministry of Youth Affairs and Sports, aimed at training young volunteers to identify local risks and support district road-safety audits in high-fatality areas.

A Long Bet on Systems

India’s road-safety response is gradually becoming more systemic, linking crash databases, analytics, driver education and enforcement into a broader operating model.

Whether that produces safer roads at scale will depend on execution. But the direction is clear: better data first, sharper intervention next, and fewer decisions made in the dark.