Every accident also means a thousand accidents that did not happen

The best use of AI is for improving safety says this art lover, innovator, and a railways veteran – who puts the common-man’s experience and safety on the same seat when he talks about future-forward contours. At the Zinnov Confluence 2025, we caught up with Dr. Sudhanshu Mani, Retired GM, Indian Railways, an Independent Consultant and an expert voice on Vande Bharat Express as the leader of the project. We tried to get some fleeting glimpses of the superfast and super-exciting coaches of technology that are being added to our train journeys. Hop on.

The very perception of a train journey has changed a lot in India, with some recent transformations. Are we on the right path?

There have been a lot of changes. Like no unmanned level crossings (that used to cause many accidents and fatalities). The modification of train toilets to bio-toilets has been a big improvement too—and that is significant because it not only created dirty sights earlier but also corrosion on tracks. But what can be done is much more than what has been done. The pace is not fast enough. Making track and fixed infra capabilities ready for 180/160 kmph train operation has not happened although we have Vande Bharat trains which are capable of operation at this speed, and the potential of these trains are therefore not fully exploited.

That would be the real value of AI in Railways. That will help us to move to a zero-accident regime.

We have heard recent announcements of incorporation of machine vision for inspection and maintenance along with use of AI in many areas in railways. What is your reckoning here?

AI is being used in many areas. Like detecting crowding through CCTV, or even in areas as seemingly ridiculous as toilet-stink-management. But my take is that the best use of AI is for improving safety. We are talking about digital record of every action at stations and in trains being available, like what signal, what time, what station, data on movement, speed, stops etc. powered through Data Loggers and on-board computers sensors. This data is available but can be collated for a real-time monitoring to detect all near misses in a better way on daily basis, followed by effective follow up (instead of them being used only for analysis after the event).

Making track and fixed infra capabilities ready for 180/160 kmph train operation has not happened although we have Vande Bharat trains capable of operation at this speed.

You mean AI’s use in preventing accidents?

Every accident also means a thousand accidents that did not happen. Those records can be so insightful for safety. The data is available, ready for meaningful exploitation. That’s where AI can step in as it is nearly impossible to analyse the data all over Indian Railways—manually. Every day there could be five-to-six near-misses and a lot of proactive strength can be cultivated as a safety culture through AI’s assistance; this post-event analysis of near misses (averted) can be so useful to prevent future ones—but crunching it all is not possible through manual approaches. That would be the real value of AI in Railways. That will help us to move towards a zero-accident regime.

Vande Bharat’s propositions mainly lean towards speed. Do speed and comfort or speed and safety get into a conflict—ever?

Not really. There are so many other reasons for an accident but the speed capacity is decided on how much a train, and the fixed infrastructure like track and signalling can handle. Speed, per se, is not that much of an issue with safety so I do not see any conflict. What we need is track and signalling system compatible with the proposed speed.

How much can IoTs and other technologies help in collision prevention and predictive maintenance?

Right now predictive maintenance is key. With on-board and on-site monitoring, some work has happened and some remains to be done. Sometimes, we also run the risk of unnecessary maintenance. Unscheduled work can hamper reliability and incur downtime. So that can be addressed in a better way.

Is digital safer than manual in transport industry? Specially after some tragic events in the transport industry that the world has gone through? And with the rise of cyber-attacks?

In my opinion, automation and digitalisation do help greatly. If you see data, flights have increased manifold and overall safety has improved. With exceptions of malfunction, usually digital is a fail-safe approach. As to cybersecurity, this area needs focus and agility because if 30 minutes of downtime on the Railways website can cause so many issues, what happens when AI comes in? The infrastructure can become amenable for attacks in any organisation where security is not put on the front-seat.

Indian Railways ranks amongst the world’s largest employers. How well do people and automation align at a scale like this? Any dream scenario you envisage?

Automation is undoubtedly desirable and you can see its penetration everywhere so it would be defeatist to downplay it. The biggest block is something I am worried about – and that’s how the common man travels in India. It’s a blot on human dignity. It’s a state of affairs that can change—if we make up our minds about it. That’s the direction I would want to go.

pratimah@cybermedia.co.in