PhD scholar Jinit Lad and researchers from the National Institute of Technology (NIT), Calicut is developing an AI-based driving licence assessment system that could evaluate motorists’ reactions to sudden hazards, decision-making and driving behaviour. The technology is still at the prototype stage. Special Arrangement
Technology

One wrong move, and AI spots it: India’s driving licence tests are set to get smarter

Ready for Driving test 2.0? AI could soon judge how safely you drive, not just how you steer. But NIT-Calicut tool still in trial stage

Shumaila Firoz

Hyderabad: Speed thrills, but kills. And most drivers on Indian roads choose to be blissfully oblivious to this cardinal rule. Jumping the signal? Fooling the cop? All a part of adreneline rush for them.

Now, you can fool humans. But what about artificial intelligence? With AI ruling every aspect of our lives, what if we say that the synthetic intellect could spot dangerous drivers BEFORE they get a licence? Or that it could potentally determine whether a person is genuinely prepared to drive on Indian roads, rather than simply capable of passing a driving test?

Researchers at the National Institute of Technology (NIT), Calicut, are developing an AI-based driving licence assessment framework that aims to do just that by evaluating how drivers react to unexpected hazards, make decisions and respond to changing road conditions.

Presenting the research at the 85th Indian Roads Congress (IRC) Annual Session at HITEX, Hyderabad, on Thursday, PhD scholar Jinit Lad explained that the proposed system would use cameras installed inside vehicles to record driving behaviour. The footage would then be analysed using artificial intelligence and machine learning tools.

Unlike conventional driving tests conducted on designated tracks, the proposed framework aims to assess aspects of road safety awareness and driver behaviour that may otherwise go unnoticed.

Beyond conventional driving tests

According to Lad, the existing driving licence testing process has several limitations, particularly when it comes to assessing hazard perception and decision-making.

Most tests are conducted in controlled environments, where the range of potential hazards is limited. As a result, candidates may be assessed primarily on their ability to manoeuvre a vehicle rather than their preparedness for unpredictable traffic situations.

The proposed system would analyse factors such as eye movement, reaction time, hazard recognition and the decisions drivers make when confronted with unexpected situations.

During a post-session interaction, a speaker involved in the discussion explained that cameras would only capture the footage, while AI tools would subsequently analyse the driver’s responses.

For instance, if an obstacle suddenly appeared in front of a vehicle, the system could examine how quickly the driver recognised it and what action followed.

The research also proposes collecting information about candidates’ previous driving experience, including supervised driving hours, alongside vision tests and AI-assisted performance evaluation.

Prototype shows promise, but challenges remain

The technology, however, is still in its early stages of development. During the interaction after the presentation, the speaker said the researchers had collected two or three initial samples, but the results had not yet met expectations.

The team is now working to identify problems, refine the system and validate its performance before moving towards wider testing.

The proposed framework also includes an educational component intended to provide candidates with feedback on their performance and help them better understand safe driving practices.

Could Telangana adopt AI-based driving tests?

Asked whether the framework could eventually be integrated into Telangana’s driving licence testing process, a speaker said the research was being developed with potential nationwide application in mind.

The discussion also touched upon developments in Gujarat, where two or three Regional Transport Offices (RTOs) were said to be involved in efforts to introduce camera-assisted monitoring on driving test tracks.

However, the speaker clarified that these were not trials of the NIT Calicut assessment framework. The existing initiatives use cameras to record driving behaviour while retaining the standard testing process.

No confirmed timeline for implementing the proposed framework in Telangana was provided.

With India recording more than 4.8 lakh road accidents and over 1.72 lakh deaths in 2023, according to figures cited during the presentation, the researchers believe that improving driver assessment and road safety education could be an important step towards safer roads.

For now, the technology remains a research project. Its wider adoption will depend on successful validation, real-world testing and acceptance by transport authorities.