A Corridor-Scale CARLA-VISSIM Co-Simulation Framework for Multi-Intersection Urban Traffic
This paper presents a stable, corridor-scale CARLA-VISSIM co-simulation framework for an urban corridor in Chattanooga, Tennessee, which integrates high-fidelity 3D rendering with microscopic traffic modeling to effectively verify cross-simulator consistency and support perception-ready studies of multi-intersection traffic dynamics.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are trying to test how a new self-driving car behaves in a busy city. You can't just put it on the real streets yet because it's too dangerous and expensive. So, you need a video game that looks exactly like the real world, but also needs to act exactly like real traffic.
This paper describes a project where the researchers built a "super-video game" by stitching together two different simulation tools to create a realistic, large-scale test track for a specific street in Chattanooga, Tennessee.
Here is how they did it, explained simply:
The Two Tools: The Director and the Actor
Think of the two software programs they used as a movie production team:
- VISSIM (The Traffic Director): This is a tool engineers use to plan traffic. It's great at the boring but crucial math: "How many cars are there?" "When should the traffic light turn green?" "How do pedestrians cross the street?" However, it looks like a flat, 2D map. It doesn't look like a real city.
- CARLA (The Visual Actor): This is a high-end video game engine (based on Unreal Engine 5). It creates beautiful, 3D, photorealistic worlds where you can see buildings, trees, and cars in 3D. It's perfect for testing how a car's "eyes" (cameras and sensors) work. But, by itself, it's not very good at managing complex traffic rules for hundreds of cars at once.
The Problem: Usually, these two tools don't talk to each other.
The Solution: The researchers built a "translator" that lets them work together in perfect sync.
The Test Track: A Real Street in a Game
They didn't just make up a fake city. They took a real stretch of Martin Luther King Jr. Boulevard in Chattanooga, which has about 15 intersections.
- They used real data (like satellite scans and 3D laser maps) to build the road so the hills and curves were exactly right.
- They imported this same road layout into both the "Director" (VISSIM) and the "Actor" (CARLA).
How They Worked Together: The "Handshake"
The researchers set up a system where the two programs run side-by-side, taking tiny steps forward in time together (like a dance where partners move in perfect lockstep).
- The Rules: VISSIM is the boss of the traffic. It decides when the lights change, where the pedestrians walk, and how the background cars move.
- The Visuals: CARLA takes those rules and draws them in 3D. If VISSIM says "The light is red," CARLA instantly turns the traffic light red in the 3D world.
- The Ego Vehicle: They added one special car (the "ego" vehicle) that is controlled by a human driver or an AI. This car lives in the 3D world (CARLA) but has to react to the traffic rules set by the Director (VISSIM).
The "Ghost" Problem and How They Solved It
When you try to run two simulations at once, a common problem is "ghosting." This happens if the Director says a car is at Point A, but the Actor thinks the car is at Point B, or if the car appears twice.
To fix this, the researchers created strict rules:
- Ownership: If VISSIM creates a pedestrian, that pedestrian belongs to VISSIM. CARLA just draws a copy of that pedestrian. If the pedestrian disappears in VISSIM, they vanish in CARLA immediately.
- The "Latest News" Rule: If the two computers send information back and forth and there's a tiny delay, the system ignores old messages. It only uses the very latest update for every car and person. This keeps everything moving smoothly without getting stuck in a traffic jam of data.
What They Found
They tested this system with about 100 cars and 100 pedestrians moving through the 15-intersection corridor.
- It Worked: The traffic lights changed at the right time in both the math world and the 3D world.
- No Glitches: The cars and pedestrians didn't glitch or disappear. When the light turned red, the cars stopped. When the "Walk" sign came on, the pedestrians crossed.
- Human Interaction: They even had a human driving the special "ego" car in the 3D world, and that human had to stop for the red lights and wait for pedestrians, just like in real life.
The Bottom Line
The researchers proved that you can take a real-world street, build a high-quality 3D version of it, and run a complex traffic simulation where a human or AI car can interact with hundreds of other vehicles and pedestrians.
They found that you don't need to simulate an entire city to get good results; simulating just one busy "corridor" (a long stretch of connected streets) is enough to test how well these systems work together. This creates a safe, realistic "playground" for testing future self-driving cars without ever leaving the computer.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.