Technical direction / systems simulation / Unreal

Smart-City Traffic + Energy Simulation

A decision system that learns traffic-light timing from simulation data, measures the tradeoff between wait time and energy demand, and replays the result as a living 3D city.

Launch the replay Watch the Unreal capture

12.9%: wait-time improvement
46.48: cars released / minute
97.6%: delivered energy ratio
1.000: decision-tree R2

The project

One logged system, three ways to see it.

The simulation models four intersections and four energy zones. Python runs the discrete-event world, trains the optimizer, and writes the evidence. AngelScript carries the data and loop concepts toward Unreal. Three.js gives the result a direct, inspectable replay surface.

The browser viewer is deliberately a replay, not a second simulation. It consumes the same logged output that produced the charts.

Observed results

Optimization with a visible trail.

The numbers below come from the verified run packaged with this showcase.

4.09soptimized average wait4.69s baseline

8.89s95th percentile waitTask 6 measurement

231.65 kWhautomated energy shed24-hour model

4intersections and energy zonesshared city layout

Traffic and energy comparison chart — Traffic flow and energy demand

Optimization landscape chart — Parameter sweep and decision boundary

Baseline and optimized wait time comparison — Baseline versus optimized wait time

Interactive artifact

Replay the optimized city.

Open replay full screen

Orbit the city, watch the signal states change, and follow the released vehicles. The viewer reads `data/light_schedule.json` and does not invent a new result.

Unreal evidence

The same idea in a world view.

A captured project view accompanies the simulation evidence.

Evidence wall

From code output to scene proof.

Smart-city scene with traffic signals and buildings — City scene and signal layout

Second Unreal smart-city scene view — World-state view

Third Unreal smart-city scene view — Runtime composition

Unreal debug board showing simulation data — Debug board and readback

Implementation

Python for the model. AngelScript for the bridge.

Browse selected source

Python

Simulate, optimize, measure.

The Python layer owns city layout, traffic arrivals, light control, energy demand, parameter sweeps, metrics, and the JSON replay contract.

View Python files

AngelScript

Carry the system into Unreal.

Small Unreal-facing examples describe the data profile and loop director without exposing the private Unreal project or its content assets.

View AngelScript files

Three.js

Make the evidence inspectable.

The viewer is a transparent presentation layer over recorded simulation output, built for direct inspection in a browser.

Open the viewer

Public boundary

A portfolio window, not the whole workshop.

The Unreal project, maps, cooked content, private plugins, and internal AYO infrastructure remain outside this public artifact. This page shows the reasoning, evidence, and result without handing over the production project.

Open the result summary