Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Traffic-Aware Optimal Taxi Placement Using Graph Neural Network-Based Reinforcement Learning

Sonia Khetarpaul, P Y Sharan

TL;DR

The paper addresses dynamic taxi placement under real-time traffic by modeling the urban road network as a graph and using Graph Neural Network embeddings to capture spatial–temporal dependencies. A Q-learning agent selects hotspot locations from a reduced set of influential nodes determined by a $k$-hop dominating set, guided by a multi-objective reward that balances waiting time, travel distance, and congestion. Experimental results on a Delhi-inspired dataset show substantial improvements in passenger wait times and driver distances compared to baselines, and demonstrate the method’s adaptability to changing traffic conditions. The approach is scalable, extends to multi-modal mobility, and can be integrated into smart city platforms for real-time urban mobility optimization.

Abstract

In the context of smart city transportation, efficient matching of taxi supply with passenger demand requires real-time integration of urban traffic network data and mobility patterns. Conventional taxi hotspot prediction models often rely solely on historical demand, overlooking dynamic influences such as traffic congestion, road incidents, and public events. This paper presents a traffic-aware, graph-based reinforcement learning (RL) framework for optimal taxi placement in metropolitan environments. The urban road network is modeled as a graph where intersections represent nodes, road segments serve as edges, and node attributes capture historical demand, event proximity, and real-time congestion scores obtained from live traffic APIs. Graph Neural Network (GNN) embeddings are employed to encode spatial-temporal dependencies within the traffic network, which are then used by a Q-learning agent to recommend optimal taxi hotspots. The reward mechanism jointly optimizes passenger waiting time, driver travel distance, and congestion avoidance. Experiments on a simulated Delhi taxi dataset, generated using real geospatial boundaries and historic ride-hailing request patterns, demonstrate that the proposed model reduced passenger waiting time by about 56% and reduced travel distance by 38% compared to baseline stochastic selection. The proposed approach is adaptable to multi-modal transport systems and can be integrated into smart city platforms for real-time urban mobility optimization.

Traffic-Aware Optimal Taxi Placement Using Graph Neural Network-Based Reinforcement Learning

TL;DR

The paper addresses dynamic taxi placement under real-time traffic by modeling the urban road network as a graph and using Graph Neural Network embeddings to capture spatial–temporal dependencies. A Q-learning agent selects hotspot locations from a reduced set of influential nodes determined by a -hop dominating set, guided by a multi-objective reward that balances waiting time, travel distance, and congestion. Experimental results on a Delhi-inspired dataset show substantial improvements in passenger wait times and driver distances compared to baselines, and demonstrate the method’s adaptability to changing traffic conditions. The approach is scalable, extends to multi-modal mobility, and can be integrated into smart city platforms for real-time urban mobility optimization.

Abstract

In the context of smart city transportation, efficient matching of taxi supply with passenger demand requires real-time integration of urban traffic network data and mobility patterns. Conventional taxi hotspot prediction models often rely solely on historical demand, overlooking dynamic influences such as traffic congestion, road incidents, and public events. This paper presents a traffic-aware, graph-based reinforcement learning (RL) framework for optimal taxi placement in metropolitan environments. The urban road network is modeled as a graph where intersections represent nodes, road segments serve as edges, and node attributes capture historical demand, event proximity, and real-time congestion scores obtained from live traffic APIs. Graph Neural Network (GNN) embeddings are employed to encode spatial-temporal dependencies within the traffic network, which are then used by a Q-learning agent to recommend optimal taxi hotspots. The reward mechanism jointly optimizes passenger waiting time, driver travel distance, and congestion avoidance. Experiments on a simulated Delhi taxi dataset, generated using real geospatial boundaries and historic ride-hailing request patterns, demonstrate that the proposed model reduced passenger waiting time by about 56% and reduced travel distance by 38% compared to baseline stochastic selection. The proposed approach is adaptable to multi-modal transport systems and can be integrated into smart city platforms for real-time urban mobility optimization.
Paper Structure (21 sections, 5 equations, 5 figures, 1 table)

This paper contains 21 sections, 5 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset and Tools
  4. Proposed GNN-Based Reinforcement Learning Framework
  5. Proposed GNN-Based Reinforcement Learning Framework
  6. Graph Construction
  7. $k$-Hop Dominating Set for Influential Node Selection
  8. GNN-Based Node Embedding
  9. Reinforcement Learning Framework
  10. State Space
  11. Action Space
  12. Reward Function
  13. Learning Algorithm
  14. Experimental Results
  15. Experimental Setup
  16. ...and 6 more sections

Figures (5)

  • Figure 1: System architecture and flow
  • Figure 2: Taxi Requests and Traffic Condition Scores
  • Figure 3: Traffic score heatmap for hotspots. Darker colors indicate higher congestion.
  • Figure 4: Traffic congestion map with top-3 recommended hotspots (red markers).
  • Figure 5: Training convergence: reward vs. travel distance.