Traffic flow dynamics: lane (de)formation

Traffic in India is notorious for being highly unruly and disorganized. Is this solely due to a lack of discipline on the part of drivers, or does the high density and heterogeneous nature of Indian traffic (mix of scooters, cars, buses and trucks) contribute to the emergence of collective chaotic dynamics? With the number of vehicles on our roads increasing each year, it is important to understand the fundamentals of Indian traffic flow in order to engineer solutions to problems like traffic congestion that plague our major cities.

Traffic flow is distinguished from other many-body problems of physics, like granular flow, by the fact that the individual agents are intelligent---drivers have memory, learned behaviours, and the freedom to violate rules. Our work aims to combine multi-agent models with reinforcement learning algorithms to understand how patterns of traffic emerge and evolve, with a focus on Indian traffic and its unique features.

This project will focus on the issue of driving in lanes. An interesting starting point is provided by the observation that while simple non-intelligent active particles spontaneously form lanes without any traffic rules.

Collaborators: Danny Raj (Danny Raj, IISc, Bangalore).

Name of Faculty