9 hours ago
IIT-Kgp app helps commuters pick ‘greener' routes on the road
Bengaluru: Ambient air pollution is responsible for 7.2% of deaths in major Indian cities every year. There's reason to believe airborne particulate matter can cut the life expectancy of Indians by up to five years.
But traffic-related pollution is usually much worse than what urban sensors report. Researchers have estimated commuting takes up only around 8% of a person's day but accounts for 33% of their pollution exposure.
IIT Kharagpur associate professor Arkopal Kishore Goswami, his PhD student Kapil Kumar Meena, and intern Aditya Kumar Singh (from IIITM Gwalior) found that while traffic significantly affects commuters' health, few were aware of its actual risks.
Realising access to information was key, the team created the Dynamic Route Planning for Urban Green Mobility (or DRUM) web app. It's like Google Maps but with the added feature of allowing users pick routes based on air quality and energy efficiency.
Cleaner commute
DRUM gives users five route options: shortest, fastest, least exposure to air pollution (LEAP), least energy consumption route (LECR), and a combination of all four factors called the suggested route.
These options are based on real-time air and traffic data. When applied to Delhi, the LEAP route reduced exposure by over 50% in Central Delhi while increasing commute time by 40%. LECR meanwhile helped reduce energy consumption by 28% in South Delhi.
These tradeoffs may not work for everyone, especially given the added fuel costs of longer routes, but DRUM could make a difference for more vulnerable groups, Mr. Meena said.
Behind the build
Integrating real-time air and traffic data was the project's biggest technical challenge, per Mr. Meena. The team's first obstacle was sparse data collection. According to UrbanEmissions, India needs around 4,000 continuous air quality stations. But by late 2024 the Central Pollution Control Board (CPCB) operated only 1,385, Mr. Meena said.
This shortfall is particularly stark in megacities like Delhi. Its 40 monitoring stations leave many areas in a blindspot.
Instead, the team relied on data from the CPCB and the World Air Quality Index. They implemented a segment-wise interpolation strategy to estimate pollution levels in areas without direct sensor coverage, divided routes into segments, and used nearby sensor data to estimate pollution where coverage was missing.
To achieve higher responsiveness, DRUM was designed to fetch live pollution and traffic data the moment a user entered a route instead of pulling data at intervals. The backend was optimised for speed while the frontend offered a clean interface.
DRUM determines routes using GraphHopper, a Java-based routing library that generates multiple options, while fetching real-time traffic updates from Mapbox. This setup allows the system to handle different vehicles and adapt to cities beyond Delhi.
How it works
At the heart of DRUM is a rank-based elimination method. 'The logic is deliberately practical: we prioritise time first because exposure is a function of concentration times time — the longer you're exposed, the more pollutants you inhale.'
Next comes distance, since shorter routes have lower emissions and fuel use even if the travel time is similar. 'After that,' Mr. Meena continued, 'we eliminate routes with higher pollution exposure, and finally, those with higher energy consumption, which we calculate based on elevation and average speed. The final output is a single suggested route that balances all four factors.'
To test the system, the team simulated Delhi's East, South, North, and Central corridors, accounting for different traffic, road quality, and pollution patterns. The results showed that shorter or faster routes often passed through polluted zones, offsetting time or distance gains.
What next?
DRUM has shown promise in simulations and Prof Goswami's MUST Lab at IIT-Kharagpur now plans real-world tests. They're also exploring integrating crowdsourced data with data from low-cost sensors on vehicles, street poles or even those carried by commuters.
'A major advantage of crowdsourced data is that it would allow us to expand the model beyond cars and two-wheelers, which are currently the only modes included,' Mr. Meena said. 'With user-contributed data from cyclists or pedestrians … we could incorporate micro-mobility modes.'
The team is also looking to DRUM 2.0, a predictive version that responds to current data as well as forecasts future air quality, traffic, and energy use. Using machine learning models such as LSTM or Prophet, it could suggest the best route now and the best time to leave. This shift would make DRUM a truly smart mobility assistant, tailored for daily life in India's most polluted cities.
Ashmita Gupta is a science writer.
