High-Fidelity Modeling of Curbside Driving Behavior
summary
This work builds a high-resolution curbside model for SUMO using the Traci interface, with detectors that track vacancy, randomized downstream spot selection from 10–80 curb slots, and a rule that allows double parking when no open slot exists. It runs 10,000 simulations on HPC and finds a clear threshold near 1500 vehicles per hour: improved behavior raises curb occupancy, edge density, and observed flow while reducing waiting time and variance compared to SUMO’s default.
AI-written summary · about