Dealing with mixed and non-normative traffic. An agent-based simulation with the GAMA platform

Area of Science:

• Computer Science
• Transportation Engineering
• Urban Planning

Background:

• Agent-based modeling frameworks are increasingly capable of city-scale traffic simulations.
• Adapting these frameworks for specific contexts (e.g., accidents, evacuations) and non-computer scientists remains challenging.
• Detailed driver behavior integration is crucial for realistic traffic simulations.

Purpose of the Study:

• To present a built-in traffic model within the GAMA platform for easier definition of agent-based traffic simulations.
• To enable detailed representation of driver operational behaviors, including infrastructure and traffic signal interactions.
• To facilitate city-level traffic simulations with a large number of agents.

Main Methods:

• Integration of a new traffic model into the GAMA open-source modeling and simulation platform.
• Development of agent behaviors to include road infrastructure, traffic signals, and lane changes.
• Implementation of support for mixed traffic (cars and motorbikes) and city-scale simulations.

Main Results:

• The model allows for the definition of complex traffic simulations with detailed driver behaviors.
• It supports city-level simulations involving tens of thousands of driver agents.
• An experiment demonstrated the model's ability to accurately replicate traffic conditions in Hanoi, Vietnam.

Conclusions:

• The developed GAMA platform model simplifies the creation of sophisticated, large-scale traffic simulations.
• It enhances the study of traffic dynamics by incorporating realistic driver behaviors and mixed traffic conditions.
• The model's accuracy in reproducing real-world traffic, as shown in the Hanoi case study, validates its utility.