IIT Guwahati researchers use Drones and AI to decode traffic patterns on Indian roads

Indian Institute of Technology Guwahati (IIT Guwahati) researchers have used high-resolution drone-based traffic data from a highway near Delhi to develop traffic stream models that can help analyse and predict traffic movement under heterogeneous Indian road conditions.

Indian Institute of Technology Guwahati (IIT Guwahati) researchers have used high-resolution drone-based traffic data from a highway near Delhi to develop traffic stream models that can help analyse and predict traffic movement under heterogeneous Indian road conditions. The study captures not only the relationship between traffic density and vehicle speed but also the variability in speeds observed under similar traffic conditions.

Led by Prof. Akhilesh K. Maurya, Professor, Department of Civil Engineering, the study tracked more than 36,000 vehicles and nearly 9 million vehicle positions on a Delhi highway. The research team used this large dataset to study how vehicle speed changes with traffic density. Based on this data, the IIT Guwahati research team developed mathematical and Machine Learning models for traffic prediction.

Traffic stream modelling uses computer-based and mathematical methods to understand how vehicles move on roads. Such models can be useful for traffic management, road design, traffic signal control, emergency response, traffic simulation, capacity assessment and the development of intelligent transportation systems. However, many existing traffic models are based on road conditions where traffic is relatively orderly, and vehicles are more homogeneous.

These conditions are different from those on Indian roads, where several types of vehicles, including two-wheelers, three-wheelers and four-wheelers, share the road along with pedestrians. This heterogeneity results in greater variability in vehicle interactions, speeds and spacing, which may not always be adequately represented by conventional traffic models. Speaking about the need for India-specific traffic data, Prof. Akhilesh Kumar Maurya said, “Traffic conditions in India are different from those in many countries where existing traffic models have been developed. We need traffic data that represents the diversity and characteristics of Indian roads. Drone-based data collection provides an opportunity to observe a large number of vehicles and their movements over a road section at high spatial and temporal resolution.”

To overcome the limitations of existing traffic stream models in Indian road conditions, the research team used drones to collect video footage from an eight-lane divided highway (wide paved shoulders) on the outskirts of Delhi during daytime traffic. The team used this data to analyse the relationship between traffic density and vehicle speed.

Based on the collected data, the research team found that: on relatively empty roads, vehicles tend to maintain larger gaps from one another as traffic becomes moderately heavy, the gaps between vehicles become smaller and vehicles tend to follow one another more closely, When traffic becomes very heavy, the gaps between vehicles increase again as traffic conditions undergo substantial changes near congested conditions. These observations indicate that the relationship between traffic concentration, vehicle spacing and speed is not necessarily linear and can change significantly as traffic moves from free-flow to congested conditions.

In addition to this, the researchers observed that before congestion set in, nearly 3,000 vehicles passed a point in each lane under full traffic flow. During traffic jams, when vehicles were not moving, the density reached nearly 900 vehicles per kilometre of road through entire width. Using these insights, the research team tested six mathematical models and trained two Machine Learning models. Considering the variability and unpredictability of traffic on Indian roads, the researchers developed the models to estimate a range of possible vehicle speeds rather than a single speed value. This approach allows the inherent variability of heterogeneous traffic to be represented more realistically.

Speaking about the real-world impact of this research, Prof. Maurya said, “Reliable traffic management begins with models that represent how traffic actually behaves on Indian roads. The high-resolution trajectory data generated through drones allows us to capture interactions among different vehicle types and the large variability that conventional traffic models often overlook. Such India-specific models can support more realistic traffic simulation, better road and intersection design, congestion management, capacity assessment and the development of intelligent traffic-management systems suited to Indian conditions.” The study highlights the potential of drone-based data collection for developing traffic models based on Indian road conditions.

The researchers believe that collecting traffic data at a larger scale and across different types of roads, cities, and traffic conditions can further improve the understanding and prediction of traffic movement in India. In future, such high-resolution traffic datasets and India-specific models can also contribute to applications such as urban traffic digital twins, adaptive traffic management, road-capacity assessment, traffic simulation and the evaluation of emerging connected and automated mobility systems under Indian traffic conditions.

The findings of this research have been published in the prestigious International Journal of Intelligent Transportation Systems Research in a paper co-authored by Prof. Akhilesh K. Maurya, along with his research scholars, Shekhar Singh and Surya H. Ravikumar. (ANI)