Crossing Simulation

A two-way road crossing simulation, running fully in the browser via WebAssembly.

Simulation

Controller

Min arrival — Way H (s)

Max arrival — Way H (s)

Min arrival — Way V (s)

Max arrival — Way V (s)

Min arrival (s)

Max arrival (s)

Green — Way H (s) 13.0s

Green — Way V (s) 13.0s

Yellow (s) 5.0s

All-red (s) 2.0s

Auto-time (TODO.md 4.6)

Way H — predicted

Way H — measured

Way V — predicted

Way V — measured

Keyboard shortcuts

Space — pause / resume
S — single-step (while paused)
R — restart with the current settings
? — toggle this help

Stats

Total spawned: 0

Elapsed: 0.0 s

Way H (East-West)

Cars on road: 0
Braking: 0
Accelerating: 0
Cruising: 0

Way V (North-South)

Cars on road: 0
Braking: 0
Accelerating: 0
Cruising: 0

Total spawned: 0

Elapsed: 0.0 s

Departed: 0

On the roundabout

Cars: 0
Braking: 0
Accelerating: 0
Cruising: 0

What the settings mean

Controller: How the crossing is managed. Semaphore is a traffic light that alternates right of way between the two roads, braking or stopping cars on red/yellow. DoNothing is a no-op controller — useful as a baseline, or to watch what happens with no traffic control at all.
Min / Max arrival (s): The range, in seconds, for how far apart two cars can arrive on the same road. Each new arrival picks a random gap inside this range — lower values mean busier traffic. For the crossing, Way H and Way V each have their own range, so one road can be made busier than the other; the roundabout has a single shared range across every approach instead.
Green / Yellow / All-red (s): Live sliders for the running Semaphore's own timing, once a crossing run with it is started - drag the all-red or yellow slider down to almost nothing and watch a car get caught mid-crossing when the other road's light turns green, then push it back up to see the danger go away. All-red is the brief phase where both roads are red at once, after yellow and before the other road gets its turn. Way H and Way V each have their own green duration.
Apply Webster timing: Computes the theoretically delay-minimizing cycle length and green split (Webster's method, 1958) from each way's measured arrival flow so far, and applies it live. Shows that prediction's own expected average delay per vehicle alongside the delay this run has actually measured (stopped time per departed car) - a rough validation of the theory against the microsimulation, not a perfect match (Webster's delay formula counts more than just stopped time). Needs a few seconds of a running crossing first, to have a real flow to measure.