HomeAlgorithms & AIMax-Pressure Traffic Signal Control — Microscopic 2D

Max-Pressure Traffic Signal Control — Microscopic 2D

Interactive 2D microscopic traffic model of a 3x3 signal grid: individual vehicles accelerate, brake and queue under real kinematics while every intersection runs the actual Max-Pressure adaptive control law. Compare it live against a Fixed-Time baseline under adjustable Poisson demand and saturation flow.

Algorithms & AI2DModerate60 FPS📱 Mobile-adapted⇄ 3D version
2d-urban-traffic-signal-optimization ↗ Open standalone

The 3D version of this sim tracks traffic as a continuous fluid — vehicle counts that flow and deplete through an ODE. This 2D companion instead simulates every vehicle individually: a point mass that measures the gap to whatever is ahead of it (a red stop line or another car) and brakes or accelerates using the real kinematic stopping-distance formula, exactly like a Gipps-style car-following model. Every intersection still runs the identical real Max-Pressure control law — computing a pressure per phase from its own queues and the queues immediately downstream and switching to whichever phase relieves the biggest imbalance — so you're comparing the same adaptive-control mathematics against a genuinely different, individually-simulated traffic mechanism. Watch cars queue bumper-to-bumper, brake for a red light and discharge nose-to-tail once it turns green, and toggle Max-Pressure against a naive Fixed-Time baseline under adjustable demand and saturation flow.

⚙ Under the hood

A 3x3 grid of intersections where every signal runs the real Max-Pressure adaptive control algorithm, but traffic is simulated microscopically -- individual vehicles accelerate, brake and queue under a real safe-stopping-distance car-following rule instead of the 3D companion's continuous fluid-queue model. Compare it live against a Fixed-Time baseline under adjustable Poisson demand and saturation flow.

traffic signalsadaptive controlcar-following modelurban algorithmsPoisson processmicroscopic simulation

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install

What did you find?

Add reproduction steps (optional)