Calibration of MITSIM and IDM car-following model based on NGSIM trajectory datasets

This paper studies the car-following behaviors of individual drivers in real traffic scenes using the trajectory datasets provided by Next Generation SIMulation (NGSIM) program. We calibrate Intelligent Driver Model (IDM) and MIcroscopic Traffic SIMulator (MITSIM) car-following models by using Genetic Algorithm (GA), with a special emphasize on MITSIM model, because there are already some nice works on the calibration of IDM model. We find that after calibration, the tracking gap errors of both models are normally below 30%. We also find that the parameter set (α⁺, β⁺, γ⁺, α⁻, β⁻, γ⁻) of MITSIM model obtained from different sampling trajectories roughly locate in a low-dimensional hyperplane rather than randomly distribute in the entire parameter space.