3D Traffic Light Green Wave: Timing Signals So Cars Never Stop

A Blender-built city street with five signals. Cars follow the Intelligent Driver Model; compare offset plans on a live time-space diagram.

Switch to 2D

Presets

Playback

Parameters

Camera

Layers

Quality

Model check

The science

What is modelled

Cars drive along a one-lane arterial with five signalised intersections. Each driver follows the Intelligent Driver Model (IDM): accelerate towards a desired speed, but keep a safe time headway to the car in front, and brake for a red or amber signal when it is comfortable to stop. The signals share one cycle length, but their start times (offsets) differ. In a green wave the offsets equal the travel time between signals at the progression speed, so a platoon driving at that speed meets a green light every time.

Equations

  • IDM acceleration: a = A [1 − (v/v0)4 − (s*/s)²], desired gap s* = s0 + vT + vΔv / (2√(A b)), with A = 1.3 m/s², comfortable braking b = 2 m/s², minimum gap s0 = 2 m.
  • Drivers anticipate the phase: a signal within 200 m acts as a stationary obstacle at the stop line only if it will not be green when the car gets there (time of arrival = distance / speed); a car that sees a freshly amber light keeps going if stopping would need more than 4 m/s².
  • Offsets: green wave offi = i (D/vp + error) mod C; simultaneous offi = 0; alternating offi = (i mod 2) C/2.
  • Free acceleration time: t(v) = (v0/2A) [atanh(u) + atan(u)], u = v/v0. Equilibrium gap: se = (s0 + vT) / √(1 − (v/v0)4).
  • Time-space diagram: each line is one car; green, amber and red bars show the signal phases at each intersection. A band of green that slopes at the progression speed is the green wave.

Checks run in the page

  • The simulated time for a car to accelerate from rest to 90 % of its desired speed is compared with the closed-form IDM expression.
  • The simulated gap behind a leader at constant speed is compared with the equilibrium-gap formula.
  • A single car driving at the progression speed is sent through the green-wave plan: it must meet no red light, make no stop, and take (N − 1) D / vp seconds between the first and last stop line.

Honest simplifications

Traffic is a single lane in one direction with identical signal timing; there is no turning traffic, no cross-street queues, no pedestrians and no lane changing. Drivers have slightly different desired speeds; the bus share only changes length and desired speed. The IDM parameters are typical textbook values, not calibrated to a particular street. The statistics (stops per car, delay) come from a seeded random arrival process, so two runs with other seeds differ by a few percent. The bar chart and the offset-error sweep run separate 700 s simulations with the same parameters.

What to try

  • Compare the four offset plans in the bar chart: the green wave gives the fewest stops, simultaneous signals the most.
  • Add an offset error of 15 s: the wave is destroyed and stops return, although the cycle and the green share are unchanged.
  • Set the progression speed far from the drivers' desired speed: the platoon spreads out and drifts out of the green band.
  • Increase the flow to 1000 vehicles per hour or cut the green share to 35 %: when demand approaches the capacity of the green share, queues form and no offset plan can remove them.
Loading model...
Loading Blender model...
Time-space diagram (vehicle paths and signal phases)
Stops per car against the offset error of the green wave
Stops per car for the four offset plans
Made with 🟠 Blender · Three.js