AI — A game changing proposition for Indian Railways?
If there’s a government enterprise that is torchbearer of modernization it is the Indian Railways (IR). One may easily come up with arguments to counter this statement. However, I have reasons to say what I am saying.
For the past three decades I have witnessed incremental changes that made rail journeys a little more convenient and comfortable every time I undertook one. Be it the interior of coaches or management of platforms or train schedules, IR has come a long way.
The biggest change and most visible evidence of modernization is the IRCTC website. Gone are the days, when one had to wait for hours at the booking counters with a token, fill a physical form, hunt for ‘exact change’ to book a ticket. Over 75% percent of bookings today happen on the IRCTC website. Yes, there are problems that surface every now and then. But by far it is way better than spending precious hours at booking counter halls.
So far so good. What’s next for railways?
For one of the largest rail networks that employs a staggering 1.2 million people and has an operating ratio of 120%, the obvious next frontier would be to boost operational efficiency and profitability.
How AI can help Railways boost operational efficiency?
Four key areas of operations stand to benefit from AI
1. Demand forecasting and capacity planning
2. Scheduling and delay management
3. Asset Management, Maintenance and Safety systems
4. Workforce optimization
Demand Forecasting and Capacity Planning
If you manage P&L of any part of the business, you know how it feels to leave the cash on the table. At times when you look at the number of waitlisted tickets you can’t help but think, how on earth could IR refuse service to so many customers every single day.
Indian Railways responds to the visible demand of waitlisted tickets by adding coaches to a train. However, the latent demand remains unrealized. It is made up of people who look at the waitlist numbers and log out without attempting a booking.
Using AI in the reservation systems, railways can dynamically open more tickets if it senses more searches for a certain route than others. It can also recommend adding direct trains with fewer stops to certain destinations based on demand.
Scheduling and Delay Management
In face of an increased demand and, hence, frequency of trains, AI can help effectively modulate traffic on a route to ensure the delays are kept to a minimum.
Moreover, the frequency of trains on a route are not a function of demand alone; it factors an additional time margin for each train on route. With delays under control the frequency can be increased to cater to the latent demand on a given route.
Although the delays are significantly contained as compared to yesteryears, the challenge isn’t fully addressed. Integrating AI for route optimization and traffic management, railways will be better positioned to diagnose the causes for delays and implement required solutions.
Asset Management, Maintenance & Safety Systems
Similar to a manufacturing assembly line, maintenance locks up the value of the equipment for a given time. Proactive maintenance helps reduce the downtime significantly. AI can help prioritize and schedule maintenance proactively based on the criticality of the requirement.
In fact, the window of time for maintenance is so narrow that the lockdown period was used to carry out tasks that were long pending. AI can help intelligently schedule the maintenance without adversely impacting the operations or having to wait for a lockdown.
USTAAD or Undergear Surveillance Through Artificial Intelligence Assisted Droid deserves a mention here. It is a robot developed by Indian Railways to check safety of under gear parts of train coaches.
Workforce Optimization
From a financial perspective, workforce optimization is one of the key areas that can drive profitability. To put things into perspective, IR spends Rs.22,000 crore annually on salaries and pensions.
The large workforce can be a boon or bane in terms of management. A boon because it can boast of ready pool of resources to deal with any exigency. A bane because managing productivity of such a large workforce can get extremely challenging.
Artificial intelligence can provide real-time insights and make timely recommendations to boost productivity. Based on data, it can suggest interventions could include cross-functional training, movement of workforce, managing of shift strength to deal with changing demand on the infrastructure.
How AI can help Railways become Profitable
Apart from the aforementioned operationali interventions like route/travel time optimization and better capacity utilization, AI can help IR turn profitable in several other ways.
Picture this: Around 25% working expenditure of railways is on its energy requirements. Spend on electricity in the year 2017–18 stood at a staggering Rs. 11,700 crores. A large part of it was due to the traction load.
While IR has put in some remarkable effort in the direction of energy conservation, a lot more can be done using AI.
With digitisation, IR has enough data on energy utilisation. The audits carried out collate and present data, but the inference by and large is drawn by humans. This is where AI can help. It can help railways optimize its energy spends by analysing vast amounts of data and establishing the consumption patterns.
Fraud/leakages of various kinds impact the bottom line of IR.
The sheer size of operations makes it difficult to spot frauds/anomalies. While booking frauds have been hogging limelight for their ingenuity, other types of fraud often go undetected.
With supervised learning, AI can learn and classify the patterns under various heads. It can detect anomalies under any vertical faster and provide real-time insights and recommendations.
Conclusion
AI is an area that IR needs to actively explore as it prepares itself for the next stage of modernization. As an organization that has huge amounts of customer data and is far from being technology-averse, it isn’t going be as big a leap. We sure can expect some major changes in the near future.
