I recently shared an idea with Google Maps. I do hope they look at it and get back to me to discuss ways to implement and test it. I am sure it adds great value to folks that regularly navigate manned level crossings on their drive to work or otherwise. Check out the full document here.
A snippet of the problem statement that I aim to address -
India, my
country, my home. We have mastered the art of the many. The concept of
Economies of Scale has always found a testing ground here, with plenty of
resources and human capital, and has often yielded extremely positive returns
for anyone that could imagine a brilliant idea and find opportunities to scale
it to fruition.
Opportunities
bring challenges. Challenges lead to new opportunities. This is the cycle that
keeps economies running. One of the challenges I face very frequently is
Trains. Well, I love trains! Who doesn’t! However, who enjoys watching a moving
train while sitting restless in their cars or bikes? Particularly when traffic
is piling up behind you? I speak of most dreaded, unexpected causes of traffic
pile up on roads! Yes, the Level Crossing!
Business
Standard reports* that there are nearly 18,000 manned and unmanned Level
Crossings across India. A majority of these crossings have adequate safety
equipment and personnel with plans to improve safety as well as eliminate
crossings using over/under-bridges. Nevertheless, one aspect of the level
crossings can never be eliminated – the traffic pile-up at the junction of the
railway and roadway, particularly if the crossing is in a densely populated
region.
This is the problem I am addressing. How helpful would it be if I was notified through Google Maps, when I plan my journey, of approximate minutes traffic would grow at level crossings? I could then already defer my travel time, or start as many minutes early to pass through the crossing before the train arrived there.
- Adds unexpected delay to planned travel time.
- Time spent at the junction increases if the train is arriving, but unusually slower than usual.
- Time spent at junction increases if the train is suddenly delayed, and the personnel is unable to pull up the gates fearing traffic pile-up on the railway track.
- Time spent at the junction is proportional to the time taken by the personnel to reopen the gates.
- Time spent at the junction is proportional to the unexpected increase in traffic density when the gates are pulled up owing to vehicles blocking each other from opposite directions, particularly because of the presence of highly erratic drivers/riders in India.
And here's the snippet of the solution that I propose -
Google Maps already uses Crowdsourcing to determine the speed of vehicles, thanks to location services enabled by drivers on their smartphones. Anonymous data transmitted to google servers enables the algorithms to determine if a higher density of vehicles are slowing down or speeding up. This, coupled with historical trends and predictive AI, enables maps to give us valuable insights into dynamic traffic conditions and calculate ETA between source and destination.
My idea is to extend this model from the roadways to the railways. A railway track can show a particular color (say a dotted green line) if there are no trains running on it. A dotted amber line can be shown, say upto 5 KM ahead of a moving train. A dotted red line can be shown at the exact location of a moving train.
So how to achieve this? A combination of principles from Google’s Crowdsourcing, Google’s historical trends and Predictive AI using Graph Neural Networks (DeepMind), and Google’s Waze and government data overlay, can be used to determine speed of a moving vehicle – trains – on a railway track. The principle can be applied to the vast network of railway tracks across a town, district, state, nation or even the entire world.
There is a clear difference between data collected and used from typical crowdsourcing, and data collected from railway travelers. Specifically –
- Threshold of expected data points will be significantly lesser. Consider a concentration of between 300 and 600 data points from location service enabled smartphones. I call this the ‘Passenger Dataset’.
- Besides considering how fast or slow the Passenger Dataset is moving, gather information about where this data set in dynamically positioned as well.
- Overlay this information with the dynamic traffic data of roadways. This would enable Maps to determine potentially exact times when trains pass coordinates intersecting with roadways.
- When the Passenger Dataset is stationary, i.e., coordinates of the data set have not changed for more than a min, the train can be considered as not moving.
- When the coordinates of the dataset have changed every min, the train can be considered as moving. The direction of the motion, coupled with the rate of change of coordinates can be used to determine how fast or slow the train is moving. Accordingly, the length of the amber and the red can be adjusted dynamically. As the speed increases, the length of red line in front of the train can be longer as compared to when it is moving slower. Similarly, the length of the amber in front of the red line can be adjusted.
- This collective information can then be coupled with the road traffic data to notify smartphones of potential delays at junctions intersecting the railways.




