In the realm of smart city transportation, two researchers from the Indian Institute of Technology Delhi, Sonia Khetarpaul and P Y Sharan, have developed a novel approach to optimize taxi placement using advanced machine learning techniques. Their work, published in the journal IEEE Transactions on Intelligent Transportation Systems, addresses the challenge of efficiently matching taxi supply with passenger demand in real-time.
The researchers propose a traffic-aware, graph-based reinforcement learning (RL) framework that considers dynamic factors such as traffic congestion, road incidents, and public events. Traditional models often rely solely on historical demand data, which can lead to inefficiencies. By modeling the urban road network as a graph, where intersections are nodes and road segments are edges, the researchers incorporate real-time traffic data and historical demand patterns to inform taxi placement decisions.
The framework employs Graph Neural Networks (GNNs) to encode spatial-temporal dependencies within the traffic network. These embeddings are then used by a Q-learning agent to recommend optimal taxi hotspots. The reward mechanism of the RL agent is designed to jointly optimize passenger waiting time, driver travel distance, and congestion avoidance. This multi-objective approach ensures that the solution is both efficient and sustainable.
Experiments conducted on a simulated Delhi taxi dataset, which was generated using real geospatial boundaries and historic ride-hailing request patterns, demonstrated significant improvements. The proposed model reduced passenger waiting time by about 56% and driver travel distance by 38% compared to baseline stochastic selection methods. These results highlight the potential of the framework to enhance urban mobility and reduce traffic congestion.
The adaptability of the proposed approach makes it suitable for integration into smart city platforms and multi-modal transport systems. By providing real-time recommendations for optimal taxi placement, the framework can help transportation authorities and ride-hailing services improve service quality and operational efficiency. This research contributes to the growing body of work on intelligent transportation systems and offers practical solutions for the energy and transportation sectors.
This article is based on research available at arXiv.