That urban intersections should follow fixed timing plans optimized offline for predictable rush periods.
Collections: Human still decides · Queue eliminated · Embodied work
Executive brief
The system made a high-frequency physical decision locally while coordinating only the information needed across neighbors.
AI value · Average travel time and wait time
Verified25.79% lower travel time and 40.64% lower wait time; projected emissions 21.48% lower
Before/after drive-through study, not randomization; emissions were modeled from fuel use rather than directly measured.
Before
Create fixed coordinated timing plans for rush periods and simple actuated control otherwise. → Apply the preset phase plan until engineers update it.
After
Recomputes green-time allocation every one to two seconds from incoming flows. → Share projected outflows with neighboring signals to form coordinated corridors.
Human boundary
Intersection agents allocate routine green time within traffic-safety constraints; public authorities retain configuration and shutdown rights.
Why it matters
That urban intersections should follow fixed timing plans optimized offline for predictable rush periods.
Before
How the work ran before the change.
Traffic engineers
Create fixed coordinated timing plans for rush periods and simple actuated control otherwise.
ControlOffline plans
Signal controller
Apply the preset phase plan until engineers update it.
ControlStatic schedule
What changed
That urban intersections should follow fixed timing plans optimized offline for predictable rush periods.
Decision rightHuman authority remains at the consequential boundary
After
How the same work runs now.
SURTRAC agent at each intersection
Recomputes green-time allocation every one to two seconds from incoming flows.
ControlSafety and phase constraints
Intersection agents
Share projected outflows with neighboring signals to form coordinated corridors.
ControlDecentralized coordination
Exception path
Engineers and conventional controller modes handle sensor, communication, emergency, and system failures.
Decision authority
Intersection agents allocate routine green time within traffic-safety constraints; public authorities retain configuration and shutdown rights.
| # | Actor | Action | Control |
|---|---|---|---|
| 01 | Traffic engineers | Create fixed coordinated timing plans for rush periods and simple actuated control otherwise. | Offline plans |
| 02 | Signal controller | Apply the preset phase plan until engineers update it. | Static schedule |
| # | Actor | Action | Control |
|---|---|---|---|
| 01 | SURTRAC agent at each intersection | Recomputes green-time allocation every one to two seconds from incoming flows. | Safety and phase constraints |
| 02 | Intersection agents | Share projected outflows with neighboring signals to form coordinated corridors. | Decentralized coordination |
Work that left the path
Human role before
Engineers authored and periodically updated fixed timing plans.
Human role after
Engineers define constraints, monitor performance, and maintain controllers while agents make routine second-by-second allocations.
AI role
Decentralized multi-agent planning for real-time signal phasing.
Coordinated fixed plans and simple actuated control in March 2012 before testing25.79% lower travel time and 40.64% lower wait time; projected emissions 21.48% lower
Before/after field runs in March and June 2012 · Twelve high-volume routes through nine intersections, nearly 30,000 weekday vehicles
Before/after drive-through study, not randomization; emissions were modeled from fuel use rather than directly measured.
Anti-pattern
Reporting modeled emissions as direct measurement or deploying without safe fallback timing.
Questions
Portability conditions
Reputation risk
low