AI for Road Safety: Data-Driven Solutions for Enhancing Road Safety in India
Abstract
The session opened with an overview of India’s nascent national road‑safety data platform (IRAD/EDAR) and the vision of the Centre of Excellence for Road Safety at IIT Madras. A “Thinnai” learning platform and a QR‑code‑driven hackathon (National Road Safety Hack‑athon 2026) were announced, inviting students and developers to prototype AI‑based safety tools. The remainder of the hour comprised a policy‑focused panel, moderated by Atul Singh, in which senior officials from the Ministry of Road Transport & Highways, Volvo, and LightMetrics discussed recent legislative reforms, the role of AI in enforcing traffic rules, data‑gap challenges, vehicle‑to‑vehicle communication, two‑wheeler safety, and the youth‑led “Sarak Suraksha Mitra” volunteer programme. The session closed with acknowledgements and an invitation to continue collaboration at IIT Madras.
Detailed Summary
Prof. Venkatesh Balasubramanian (Centre of Excellence for Road Safety, IIT Madras) opened the session by linking road‑safety challenges to the broader Vikshit Bharat 2047 vision of technology‑led public‑good impact. He highlighted that the Ministry of Road Transport & Highways (MORTH) had adopted a data‑first approach, championing the creation of a national road‑safety database (referred to as IRAD and EDAR) managed by the National Informatics Centre (NIC).
Key points:
- Scale of the problem – India’s population (~1.4 billion), thousands of police stations, and a massive, under‑reported crash burden.
- Uniqueness – No other country has attempted a single umbrella database covering all crash‑related variables.
- Strategic intent – The database is intended to be the “foundation layer” on which AI applications can be built (risk prediction, enforcement support, hyper‑local interventions).
Prof. Balasubramanian concluded with a call for human‑centred empathy in any technology solution and emphasized youth‑engagement as a catalyst for fresh ideas.
2. “Thinnai” Learning Platform & QR‑Code Demo
Atul Singh (COORS, IIT Madras) took the stage next. He described a three‑gate educational model:
- Knowledge Gate – Mechanisms for better comprehension of road‑safety concepts.
- Skill Gate – Structured practice opportunities.
- Practice Gate – Real‑world application and feedback.
To make the model culturally resonant, the platform is branded “Thinnai” (a traditional South‑Indian courtyard where elders teach youths). The platform, still in beta, is accessible via a QR code displayed on the screen. Attendees were asked to scan the code, explore the prototype, and provide feedback before its formal launch.
Announcements
- The QR‑code‑based demo is a call‑to‑action for participants to co‑create the next big road‑safety tool.
- The Centre invites students, developers, and innovators to join the upcoming National Road Safety Hackathon 2026, with the first 50 registrants receiving mentorship directly from IIT Madras faculty and industry partners.
3. National Road Safety Hackathon 2026 – Challenge Statements
During the hackathon launch, Atul Singh posed scenario‑based prompts to the audience (primarily Gen‑Z participants):
| Scenario | Desired AI/Tech Solution (audience ideas) |
|---|---|
| Accident near campus during Valentine’s week – need to locate a victim quickly. | Fall‑detection app, image‑based reporting, real‑time alert. |
| Need to identify the nearest suitable hospital for a brain‑injury victim. | Geo‑coded hospital database, AI triage routing. |
| Personal experience of an accident where the driver could not locate the nearest police station or appropriate hospital. | Integrated navigation + emergency‑call feature. |
These prompts were framed as problem statements for hackathon participants to address, reinforcing the need for hyper‑local, AI‑driven interventions that operate under time‑critical conditions.
4. Panel Discussion – “Policies Enabling Innovation in Road Safety”
Moderator: Atul Singh (COERS, IIT Madras)
Panelists:
- Shri Ramendra Pratap Shukla – Director, Road Safety Cell, MORTH
- Shri Pankaj Aggarwal – Chief Engineer, MORTH
- Shri Dheepan Raja Jayabalan – Volvo Group (industry perspective)
- Dr Pushkar Patwardhan – LightMetrics Technologies (technology partner)
4.1 Legislative & Institutional Foundations
- Motor Vehicle Act (MVA) 2019 – Established electronic monitoring & enforcement as a legal backbone.
- Data‑Sharing Policy – Recent directive from MORTH to make crash‑related data openly accessible to researchers and innovators.
- State‑level implementation – Enforcement tools are largely state‑run; central agency provides grants, technical assistance, and standards.
Key Insight: “AI can only be as effective as the data it receives; therefore, a unified, high‑quality data pipeline is essential.” – Shri Aggarwal
4.2 Data Gaps & Under‑Reporting
- Official estimates cite ≈150,000 road‑fatalities per year, but under‑reporting is severe (e.g., in Bihar, reported deaths ≈95 % of accidents, suggesting many incidents go undocumented).
- Program: Data‑Driven Hyperlocal Intervention – field perception surveys to capture unreported crashes.
Recommendation: Build AI models that triangulate multiple sources (police records, mobile‑phone data, crowdsourced reports) to fill gaps. – Shri Shukla
4.3 Vehicle‑to‑Vehicle (V2V) Communication
- Mandate: V2V communication frequencies allocated (30 MHz band).
- AI‑driven feedback loops to drivers (e.g., speed warnings, lane‑departure alerts) before a collision occurs.
- Challenge: Translating raw sensor data into actionable, driver‑centric messages without causing distraction.
Open Question: How to ensure interoperability across heterogeneous vehicle manufacturers? – Shri Aggarwal
4.4 Road‑Safety Education & “Thin‑AI”
- Current licensing focuses on procedural compliance; driving is treated as a life skill with no school‑based curriculum.
- Proposal: Integrate AI‑powered road‑safety modules into school curricula (starting at age 16), with special emphasis on girls and two‑wheeler safety.
Policy Suggestion: Make a “Driving Literacy” certificate a prerequisite for high‑school graduation. – Prof. Balasubramanian (cited by panel)
4.5 Two‑Wheeler & Pedestrian Safety
- Two‑wheelers account for ≈45 % of road‑fatalities; pedestrians ≈20 %.
- Existing safety features are largely car‑centric; need sensor‑based blind‑spot detection, automatic braking for two‑wheelers.
- Industry (Volvo) is exploring lightweight, affordable ADAS for motorcycles.
Key Insight: “Design safety for the vehicle type that kills the most, not the one that sells the most.” – Shri Jayabalan
4.6 “Sarak Suraksha Mitra” Volunteer Programme
- Collaboration: Ministry of Youth Affairs & Sports + MORTH.
- Goal: Deploy local youth volunteers (trained 15 days) as road‑safety auditors in the top 100 high‑fatality districts, later scaling pan‑India.
- Volunteers assist district road‑safety committees, conduct ground‑truth audits, and serve as community ambassadors for helmet use, safe‑driving habits, and data collection.
Call to Action: “Every road user—driver, passenger, pedestrian, even a Volvo bus operator—must adopt an empathetic, safety‑first mindset.” – Shri Shukla
4.7 Technology Limits & Human Factor
- Panel acknowledged that AI cannot replace human judgment; empathy, local context, and community engagement remain crucial.
- Emphasis on transparent AI: explainable alerts, audit trails to avoid misuse.
5. Closing Remarks & Acknowledgements
- Prof. Venkatesh Balasubramanian thanked the audience, reaffirmed IIT Madras’s role as an open hub for research, and invited participants to visit the campus.
- Atul Singh thanked the panelists and participants, reminded attendees to scan the QR code for the hackathon, and encouraged continued collaboration.
- Himani Suri (Centre of Excellence) delivered final thanks on behalf of IIT Madras, noting that the session marked a milestone toward scaling AI‑driven road safety across the Global South.
Key Takeaways
- National Data Infrastructure: India is building its first unified road‑safety database (IRAD/EDAR), the backbone for any AI‑driven safety solution.
- Thinnai Platform: A culturally tailored, three‑gate learning environment (knowledge, skill, practice) launched in beta, accessible via QR code.
- Hackathon 2026: Open call for AI‑based solutions to real‑world crash scenarios; first 50 teams receive IIT‑Madras mentorship.
- Legislative Enablement: The 2019 Motor Vehicle Act and MORTH’s data‑sharing policy provide legal scaffolding for electronic enforcement and AI analytics.
- Data Gaps: Severe under‑reporting, especially in states like Bihar, demands hyper‑local, crowd‑sourced data collection methods.
- V2V Communication: Mandated frequency allocation opens pathways for AI‑driven driver feedback, but standards and interoperability remain open challenges.
- Education Gap: Road‑safety literacy is absent from school curricula; integrating AI‑based modules could create a generation of safer drivers, with special emphasis on two‑wheelers and female riders.
- Two‑Wheeler Focus: With almost half of fatalities involving two‑wheelers, bespoke ADAS and policy interventions are urgently needed.
- Sarak Suraksha Mitra: Youth‑volunteer auditors will bridge the gap between policy and ground realities, creating a scalable model for local safety audits.
- Human‑AI Symbiosis: AI tools must be designed with empathy and transparency, complementing—not replacing—human judgment and community engagement.
See Also: