Smart Streetlight Grid: Adaptive Dimming Network (2D)

Network Conditions

Traffic density 45%

PIR detection radius 9 m

Idle dim level 15%

Hold time after presence 6.0 s

Live Network Readout

Power draw

0.00 kW

Energy saved

Lit at full

0 / 49

Agents on grid

How it works

Each of the 49 poles runs an independent PIR-style presence sensor and a per-light brightness state b ∈ [0,1] that relaxes toward a target with first-order dynamics:

db/dt = (target − b) / τ
b(t+dt) = b + (target − b)·(1 − e^(−dt/τ))

τ is short while rising toward a detection (≈0.35 s, an LED driver responds almost instantly) and longer while decaying back down (≈2.5 s, so the street doesn't flicker to black). The target itself follows a simple state machine per pole:

if agent within detection radius:  holdTimer ← holdTime
else:                               holdTimer -= dt

target = 1.0                         if holdTimer > 0
       = bridgeLevel (0.55)          if a 4-neighbor pole has holdTimer > 0
       = idleDim                     otherwise

The neighbor-bridging rule is what turns 49 independent sensors into a network: a pedestrian tripping one pole's sensor also half-lights the adjacent poles, so a person always walks inside a lit corridor instead of a single hard-edged bright spot in a dark street.

Power model: each fixture draws P = P_idle + (P_max − P_idle)·b, with P_max = 150 W and a non-zero P_idle = 25 W floor (the driver circuitry itself never fully switches off). Live draw sums all 49 fixtures; energy saved compares that sum against the "always at P_max" baseline that a non-adaptive streetlight network would draw:

saved % = 100 · (1 − ΣP_i / (49 · P_max))

Click and drag on the street grid to walk a manual pedestrian through the network and watch the safety corridor form around them. "Simulate sensor blackout" forces every pole to idleDim regardless of traffic, showing the safety cost of a dead network.

Click + drag on the street to walk a manual pedestrian

Full brightness Idle dim Vehicle / pedestrian Manual pedestrian