TL;DR
This paper introduces a novel game theory-based traffic signal control system that optimizes intersection flow by considering both vehicles and pedestrians simultaneously. By employing a Nash bargaining solution, the system dynamically adjusts green light durations to minimize and equalize queues for all road users, leading to significant improvements in traffic efficiency and fairness compared to traditional methods.
What Problem Does This Paper Address?
Traffic congestion is a growing problem, especially in urban areas, with intersections being major bottlenecks. While many traffic signal control strategies exist, most focus solely on vehicular traffic, neglecting pedestrians who are also crucial participants at intersections. This oversight can lead to suboptimal signal timings, increased delays, and reduced overall efficiency. The paper aims to bridge this gap by developing a signal control approach that integrates both vehicle and pedestrian needs, recognizing their potential conflicts and the need for equitable treatment. The challenge lies in creating a dynamic system that can adapt to varying traffic demands and ensure fairness across all road users.
Methods
The core of this research is a game theory-based approach, specifically utilizing the Nash Bargaining (NB) solution. The intersection is modeled as a four-player game, where each of the four phases represents a player. The "strategies" for these players are the green light durations for each phase, constrained within minimum and maximum limits. The "payoff function" for each player is defined as the negative of the estimated weighted sum of "people" (vehicles and pedestrians) waiting at the intersection, aiming to minimize queues and equalize them across phases.
A key innovation is the inclusion of pedestrians in the payoff function, using a conversion factor to account for average vehicle occupancy. The disagreement point, representing the minimum acceptable payoff for each player, is carefully calculated based on the sequential nature of the phases. The NB solution is then used to find the optimal green durations that maximize the product of the players' gains over their disagreement points. This approach is implemented and tested using the SUMO (Simulation of Urban Mobility) traffic simulator, with a connected environment assumption where vehicle data is available via VANETs and pedestrian presence is detected by cameras.
Datasets
The paper does not use pre-existing datasets for training. Instead, it relies on simulations conducted using the SUMO software. The traffic demand scenarios are generated within SUMO by varying probabilities of vehicle and pedestrian arrivals on main and branch roads. This allows for controlled experiments to evaluate the proposed algorithm under various balanced and unbalanced demand conditions, including stochastic elements.
Key Insights and Ideas
- Pedestrians are not an afterthought: The paper's most significant contribution is the explicit integration of pedestrians into the game-theoretic framework for signal control. This moves beyond traditional vehicle-centric approaches.
- Fairness through equalization: The Nash Bargaining solution, by aiming to equalize payoffs (or minimize and equalize queues), inherently promotes fairness among different traffic streams and user types. This is a crucial aspect often overlooked in purely efficiency-driven models.
- Connected environment as an enabler: The assumption of a connected environment (V2V and V2I communication) is critical for real-time data acquisition, enabling the dynamic nature of the proposed NB approach.
- Stochasticity matters: The inclusion of a random parameter "c" in traffic generation acknowledges the inherent variability in real-world traffic, making the simulations more realistic.
- Right-turning vehicles' unique challenge: The sensitivity analysis highlights the complex interactions between right-turning vehicles, pedestrians, and other traffic streams, demonstrating the NB approach's ability to manage these nuanced conflicts.
Key Results
The proposed Nash Bargaining-based signal control approach consistently outperformed the benchmark actuated (Act) signal control strategy across various metrics:
- Reduced Delays: Significant reductions in average pedestrian delay (APD) were observed, with reductions up to 17.94% in balanced scenarios and up to 31.15% in unbalanced scenarios for average vehicle delay (AVD).
- Shorter Queues: Average queue lengths (AQL) were reduced by up to 20.83% in balanced and 32.28% in unbalanced scenarios.
- Environmental Benefits: Reductions in CO2 emissions (ACE) and fuel consumption (AFC) were also achieved, ranging from 1.25% to 8.92% in balanced scenarios and 11.80% to 16.22% in unbalanced scenarios.
- Enhanced Stability and Fairness: The NB approach demonstrated more stable queue lengths with lower standard deviations, indicating a more predictable and fair distribution of traffic flow across lanes. This was particularly evident in the balanced demand scenarios.
- Right-Turning Vehicle Performance: In scenarios with high demand for right-turning vehicles, the NB approach showed substantial reductions in delay and queue length, outperforming the actuated strategy.
Limitations and Future Work
The authors acknowledge that the consideration of stochastic traffic demand in this paper is insufficient. The current model may not fully capture the highly stochastic nature of real-world traffic. Future research directions include:
- Exploring more realistic stochastic traffic demand models: To better reflect actual traffic conditions.
- Utilizing real-world data: To validate the simulation results and further refine the model.
- Extending to multi-intersection networks: The current study focuses on isolated intersections.
- Investigating adaptive cycle lengths: The current model uses a fixed cycle length.
Summary updated