Traffic Simulation Part 1: Introduction

What is Traffic Simulation?

OpenStreetMap road network data for Auburn, Alabama visualized

We probably don’t often consider the research and development that is done to construct the roadways we use each and every day. However, we probably all dislike being stuck in traffic jams or being forced to alter our preferred route to a destination to deal with congestion or other unfavorable traffic conditions. We may not always be aware of it, but the design and construction of the roadways we drive requires vast amounts of research and planning. One of the critical components of the effort to create and maintain performant roadways is simulation of traffic using modeling techniques.

Traffic Models

Traffic models are grouped into three major categories based on the strategy employed:

• Macroscopic - Uses volume, speed, and density parameters to describe flow with respect to time and space [1]

• Microscopic - Uses the actions of each vehicle (acceleration, deceleration, lane changes, etc.) to model traffic flow [1]

• Mesoscopic - Combines the dynamics of Microscopic modeling with some simplifications of Macroscopic modeling to achieve a mixture that provides better computational feasibility than pure Microscopic but still captures more detail than Macroscopic [1]

  
  

Dynamic Network Loading

Dynamic network loading is the process of placing differing loads across a network that changes over time. This can be done in two different ways: using analytical methods or heuristic based algorithms. Dynamic network loading models can be described as models that “simulate how the time-varying continuous path flows propagate through the network inducing time-varying inflows, outflows and link occupancies” [4]. Computer simulation can be used as an alternative to analytical models [1].

Traffic Simulation at Autonoma

Here at Autonoma, we primarily employ microscopic simulation strategies using heuristic algorithms for dynamic network loading. One of our strengths is high fidelity physics modeling for vehicle dynamics. We simulate a variety of parameters such as brake temperature, tire dynamics, and suspension models that allow us to accurately estimate the behavior of a vehicle on the whole given a particular environment.

With this in mind, we utilize microscopic traffic modeling techniques to provide the highest level of detail with respect to individual traffic agents. This enables us to produce high quality data for not only traffic flow given a traffic network, but also understand the potential safety concerns with respect to the frequency and severity of traffic accidents, among other critical insights about a road way.

Visualization of microscopic traffic simulation using Autonoma’s proprietary traffic simulation software

In the simplified rendering shown above, the estimated car following parameters are shown. The classical car following model is used along with additional, more complex models. We use heuristic based algorithms that estimate the agents route choice decision making first, then emulating the dynamic network loading.

One concern with high fidelity, microscopic simulation is the compute required. We utilize cutting edge vertical scalability to parallelize our simulation technology in order to mitigate potential performance bottlenecks for large traffic networks with many traffic agents.

Highly Automated and Autonomous Agents

The advent of highly automated and autonomous agents presents a unique set of challenges and opportunities. Our software allows for autonomous agents to be inserted into the traffic simulation in a way that allows for testing and validation of both the agent and the road system when autonomy of varying degrees is present. Because our software provides perception capabilities to traffic agents, autonomous systems can be tested within our software utilizing simulated sensor data such as camera, LiDAR, GPS, and inertial data.

Next Up

As this engineering blog series continues, we’ll discuss more details related to how we approach traffic simulation at Autonoma and how our expertise and experience in vehicle dynamics coupled with our technical depth in software engineering can provide capabilities that other simulation software stacks lack.

References

1. Jaume Barceló (2010). Fundamentals of Traffic Simulation. Springer New York Dordrecht Heidelberg London.

2. OpenStreetMap (2025). www.openstreetmap.org.

3. Nurul Nasuha Nor Azlan, Munzilah Md Rohani (2018). Overview Of Application Of Traffic Simulation Model. MATEC Web of Conferences 150. MUCET 2017.

4. Cascetta E (2001) Transportation systems engineering: theory and methods. Kluwer, Academic Publishers. The Netherlands