NSIT's Artificial Intelligence System Likely to Help Prevent Waterlogging in Indan Cities

A team of engineering students has developed an artificial intelligence (AI) system that can identify areas prone to waterlogging, and may help metro cities avoid tedious road congestions caused by monsoon showers. The researchers, including Aman Bansal and Apoorva Gupta from Netaji Subhas Institute of Technology (NSIT) in New Delhi, combined rainfall, traffic, and location data to predict the severity of water logging in the vulnerable areas. "In a lot of developing countries, including India, the issue of waterlogging is persistent. In 2016, the roads of Gurgaon were flooded leading to severely waterlogged streets, which left thousands of people stranded for several hours. In Mumbai too, such incidents are very common," Rishab Gupta from NSIT.

"The damages caused by such occurrences triggered us to find a practical and feasible solution," said Gupta, one of the lead authors of the study that was presented at the Institute of Electricals and Electronics Engineers' (IEEE) 15th International Conference on Smart City in Bangkok last year. The study was initially done on Manila, the capital city of Philippines which has similar environmental conditions like cities in India.

"We first marked the areas in Manila already established to be prone to waterlogging, based on data obtained from previously conducted surveys," Gupta said.

The areas susceptible to waterlogging were located with the help of past travel time data (the time taken to travel from one point to the other) sourced from smartphone-based cab service Uber, and elevation data of the area.

The intensity of waterlogging was calculated based on the rainfall data and the day of the week, as traffic on weekends is significantly less than that on weekdays.

"After we ran this data through our trained neural network, we verified the vulnerability of those locations and even came up with more areas defenceless against waterlogging," he said.

The data was fed into an artificial intelligence system that consists of a neural network that can derive patterns in the information fed to it.

The system was trained by the students to reveal a water logging intensity score using an algorithm which could determine the extremity of the problem in the area.

"Our work can be easily modulated for Delhi as the Uber data for Delhi is available now. I can produce the results in 5-10 days," said Gupta.

"This system can be trained to detect patterns on an hourly basis and for the future as well, which is something that even Google can't calculate," Gupta said. Previously researchers used the Internet of Things (IoTs) - a network of small electronic devices installed in different locations to gather data about moisture, traffic, etc. However, the high cost of these devices made the project economically unviable.

Previously researchers used the Internet of Things (IoTs) - a network of small electronic devices installed in different locations to gather data about moisture, traffic, etc. However, the high cost of these devices made the project economically devices installed in different locations to gather data about moisture, traffic, etc. However, the high cost of these devices made the project economically unviable.

"The fact that we could easily access travel time data for free from Uber which is like an open repository of traffic data made things simpler and feasible for us, and gave a practical appeal to the technology," Gupta said.

The system can also be used for pinpointing accident-prone areas and times in a city, for dynamically deciding strategic points for positioning of ambulances, calculating the effect of festivals and holidays on traffic, and can also be employed in urban road planning, he said.