Train journeys in India have been romanticized to no end
NLP can be used to make sure people can book their tickets speaking in their language on phone
AI can also help with better crowd management on the platforms
It’s been a couple of years since we have been hearing of Indian Railways planning to adopt AI. Most of the conversation is centred around the use of facial recognition to nab people with criminal antecedents.
Notably, the more obvious and immediate problems that AI can solve for the railways have been conspicuously missing from the discussion.
However, the government seems to be enthusiastic about the prospect AI offers. In a recent AI summit organized by Intel India, Union Minister Mr Ravi Shankar Prasad said, “Health, diagnostics, agriculture and education are some frontiers where AI should bring improvements in the lives of the marginalized and underprivileged.”
In this article, we will discuss AI-based solutions that can make life easier for millions of rail travellers. For travel and AI enthusiasts, there could be no better way to make up for months of being cooped up at home.
Let’s dive in straight.
Train journeys in India have been romanticized to no end. Although I yearn for one myself, in this article I will focus on the parts that inconvenience rail users and can be improved using AI.
For the sake of understanding, let’s divide the passenger experience into three parts:
- Pre-boarding: It includes journey planning and booking, commute to the station, waiting on the platform and boarding the train
- Onboard: Travel experience after boarding until deboarding the train
- Post-De-boarding: Stepping foot out of the train at destination station to reaching the individual destination (place of stay/home)
AI In Pre-boarding Experience
While enough has been written about the scams involved in booking, less is said about the challenges that a user faces when planning the journey.
Come to think of it, the booking process of IRCTC requires you to be educated and conversant with technology, which the majority of the population isn’t.
For starters, NLP can be used to make sure people can book their tickets speaking in their language on phone. Given the number of regional languages, dialects and heavy mother tongue influence, it may sound ambitious; but that’s where AI helps.
As a digitized booking platform, the IRCTC website and the app are capable of much more.
For example, based on the purpose of the journey, final destination (after deboarding), probability of delay and other factors, railways can recommend the best possible train to the user while booking. Say, a user visiting Bengaluru on business has a meeting at an office on MG Road, IRCTC can recommend deboarding at Bengaluru Cantonment and a train that halts there at 8.30 AM, more than 85% of the time.
Another aspect of the preboarding experience, one that causes much inconvenience to the elderly, is the platform for boarding. If a higher percentage of elderly or pregnant women undertake a journey, train boarding should be on a platform with easy access to such people. AI can decide the platform of arrival/departure, based on the passenger data of multiple trains arriving or departing from a station around a given time.
AI can also help with better crowd management on the platforms. Using image recognition, it can alert the authorities if the crowd exceeds the number of tickets issued – both for platform and journey. Appropriate interventions can then be made.
AI In Onboard Experience
The view outside the window, the hawkers on the platform and other nostalgic references notwithstanding, train journey brings with it its own share of anxieties and inconvenience.
Most common issues include allocation of berths in different coaches, missing early morning destinations, and of course, safety concerns arising out of travellers without tickets claiming their right on your berth.
While a few solutions like destination alerts are being implemented by the railways, they are not automated yet.
With AI at work, most of these problems can be accurately diagnosed and patterns established. Some problems of systemic and cultural origins like unauthorized travellers, hawkers entering the train appear to be unsolvable. However, AI could help figure patterns and manage the problem if not completely solve it.
For example, image recognition systems can be used to identify source stations and calculate resulting revenue losses. Backed by facial recognition evidence, the quantum of punishment/penalty for ticketless travel can be disproportionately increased at select stretches to deter such behaviour.
AI In Post-Deboarding Experience
The experience of first-time visitors at major stations can be fraught with confusion, chaos and harassment by taxi/rickshaw drivers.
IRCTC, with tons of user data, can deploy AI to ensure the passengers reach their final destination comfortably after deboarding. Booking a cab and hotel are passe. Knowing the age, final destination of the passengers, and amount of luggage the IRCTC app can automate post-travel arrangements, with minimal or no human effort, for a price included in the ticket.
Several on-demand services like assigning porter for luggage, booking transport through metro or cabs, can be provided to the user based on their booking information.
Post-de-boarding services don’t just make travel convenient for passengers but also add revenue channels to Indian Railways, which is in poor financial health (its pre-COVID operating ratio was 120%).
None of the solutions discussed is too far-fetched. Some of the applications are already used by private players in travel and other sectors. If the IRCTC site is a reality today for 75% of people who travel by train, a layer of AI sitting on top of it is hardly a tech challenge of any enormity. It is an opportunity to make one of the biggest rail networks world-class, efficient and profitable.
Further, AI can also lend a semblance of structure and order to railway stations that are symbolic and also a breeding ground of societal entropy today.