Traffic State Estimation for Prediction

To meaningfully predict what traffic might look like under different response scenarios, it is essential to have a representation of current traffic conditions in the corridor that can be used as a starting point for a simulation.

Freeway estimation

Real-time data exist at specific points along the freeway, but some gaps remain. Estimation fills in the blanks to provide a complete picture of traffic state:

The basic dimensions of freeway traffic estimation are:

Goal
Input
Output
  • Provide a complete picture of traffic conditions along a freeway based on observed data
  • Network of roads represented as links and nodes
  • Fundamental diagrams for each link
  • Boundary flows at edges of network
  • Turning movements (split ratios) at each node
  • Real-time flows and occupancies from detectors
  • Velocities and densities on each link

The estimation process looks like this:

The network description, fundamental diagrams, boundary flows, and turning movements are input into a CTM-based model and then assimilated with real-time detector data to generate the estimated freeway traffic state. (In AMS Phase 2, CTM is used for freeway estimation only, not for simulation.)

Arterial estimation

For arterial traffic estimation, the basic dimensions are:

Goal
Input
Output of current process
  • Estimation of traffic conditions on arterial segments at a given time based on observed data
  • Intersection geometry
  • Signal timing plans
  • Historical approach flows and turning counts
  • Real-time sensor counts and occupancies from advance and stop line detectors
  • Average queue lengths for each turning movement at individual intersections

Further details on the estimation process and calibration of the freeway and arterial estimators can be found in the AMS Phase 2 presentation.