Intelligent Driver Model on Straight Road

The well-known Intelligent Driver Model (IDM) proposed by Martin Treiber et al (https://arxiv.org/abs/cond-mat/0002177).

• straight road
• bottleneck
• stop-and-go waves
• stochasticity

Versions and Observations

Version 3 (v3): Stochastic inflow

Even though the inflow rate is the same, I found that its stochasticity has a substantial impact on the stop-and-go waves. Take the following figures as examples: although the inflow is consistently around 1430~1440 veh/h, the resulting stop-and-go patterns differ substantially.

Experiment 1: Deterministic Inflow (Short-lasting Bottleneck)	Experiment 2: Stochastic Inflow (Short-lasting Bottleneck)
Experiment 3: Deterministic Inflow (Long-lasting Bottleneck)	Experiment 4: Stochastic Inflow (Long-lasting Bottleneck)

Version 2 (v2): Bottleneck

• Bottleneck: At the end of the road and during a given period, a mandatory deceleration segment (bottleneck_speed_limit) was introduced to a part of vehicles (percentage_influenced_by_bottleneck) as a bottleneck, which triggered stop-and-go traffic.
• Driving noise: A random noise was added to (the driver's perception of) relative speed. IDM itself seems too stable, and noise must be added to trigger stop-and-go. The problem is: if the noise is too large, traffic breaks down itself before arriving at the bottleneck.
• Note 1: Additional constraints must be incorporated into the IDM to prevent collisions. Otherwise, collision will occur.
• Note 2: The shape of the stop-and-go is highly influenced by the values of those IDM parameters. Sometimes, very sensitive. Many trial-and-error attempts are needed. It is also unclear if the stop-and-go pattern remains the same if traffic conditions are changed.
• Note 3: It seems also difficult to set a set of values which can perfectly make the wave speed between -20 and -10 km/h. Sometimes, it is too fast (-25km/h). When trying to slow it down, waves disappeared.

Version 1 (v1): Preliminary framework

Citation

If you find this work useful, please consider citing the project

@misc{ZhengbingHe2025,
  title={Intelligent Driver Model and Stop-and-Go Waves: Code and Experiements},
  author={He, Zhengbing},
  journal={},
  howpublished = {\url{https://github.com/gotrafficgo/idm_and_traffic_wave}},
  year={2025}
}