Seminar: Discrete Event Traffic Simulation Framework using Object-Driven Cellular Automata

SEMINAR 

DEPARTMENT OF INDUSTRIAL ENGINEERING 

Discrete Event Traffic Simulation Framework using Object-Driven Cellular Automata 

Kerem Demirtaş,

Invent.AI

Abstract:

Traffic simulation frameworks face a fundamental tension: microscopic simulators like SUMO and Vissim offer behavioral fidelity but rely on fixed-timestep updates that waste computation on non-interacting vehicles, while cellular automata models are efficient but constrain speeds to integers and enforce synchronous parallel updates. Both paradigms force all vehicles to march to a global clock, even though real traffic interactions are inherently asynchronous. We propose Object-Driven Cellular Automata (ODCA), framework that resolves this tension by embedding a cellular spatial structure within a Discrete Event Simulation environment. Each vehicle operates as an independent process that acquires and releases cell resources as it traverses the freeway. Congestion is not imposed by rules — it emerges naturally when a vehicle requests an occupied cell and must wait for the resource to become available. A hybrid discretization scheme preserves the simplicity of discrete cells while allowing continuous speeds and event-driven time advancement, eliminating the artificial speed plateaus of classical CA models. The framework's architecture cleanly separates the driver process — where vehicles evaluate speed through car-following models and lane-change decisions through logistic probability functions — from the movement process, where resource acquisition governs physical progression. This decoupling makes it straightforward to swap behavioral models or introduce new vehicle types. Notably, the simulation produces output natively in the passage-time coordinate T(x,n), directly aligned with Newell's kinematic wave theory. We demonstrate the framework on a four-lane freeway network with on-ramps and off-ramps under mixed traffic scenarios with varying penetration rates of autonomous vehicles. Results show realistic fundamental diagrams, emergent congestion patterns from resource contention, and clear differentiation between human-driven and autonomous vehicle populations through their distinct behavioral parameters.

Keywords: Traffic Systems, Autonomous Vehicles, Discrete Event Simulation

Short Bio: 

Kerem Demirtaş is a Senior Data Scientist at Invent.AI. He received his B.Sc. and M.Sc. degrees in Industrial Engineering from Middle East Technical University and his Ph.D. from Arizona State University. His research interests include traffic flow modeling, network flows, simulation, supply chain management, and optimization. Before joining Invent.AI, he worked at Intel Corporation, Smart Kiwi, and Spyke Games. Outside of his professional pursuits, he enjoys real-time strategy and puzzle-oriented video games, and loves taking long walks with his dog, Gandalf.

All interested are cordially invited.  

DATE:  March 27, 2026

TIME:  Friday, 15:00

ROOM: Engineering Building, M3100