⚡ electrical engineering

Constrained Control of PDE Traffic Flow via Spatial Control Barrier Functions

This paper extends Control Lyapunov Function theory to macroscopic PDE traffic models by unifying it with spatially varying Control Barrier Functions to develop a constrained variable speed limit control strategy that stabilizes traffic density to a desired profile while strictly enforcing safety constraints.

Original authors: Brian Block, Stephanie Stockar

Published 2026-06-03

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Brian Block, Stephanie Stockar

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 a long, circular highway where cars are flowing like water in a river. Sometimes the water flows smoothly, but other times it gets too crowded, creating a "traffic jam" wave that moves backward against the flow of cars. The goal of this research is to figure out how to manage this traffic river so it stays calm and follows a specific pattern, without ever getting so crowded that it becomes dangerous.

Here is how the authors solved this problem, broken down into simple concepts:

1. The Problem: Steering the River

Traffic engineers use math (specifically Partial Differential Equations, or PDEs) to predict how traffic density changes over time and space. They want to use Variable Speed Limits (VSL)—changing the speed limit signs on the road—to control the flow.

Think of the speed limit signs as a remote control for the traffic river. If you lower the speed limit, cars slow down, and the "river" of traffic changes its shape. The challenge is twofold:

Goal A (Stability): Get the traffic to match a specific, ideal pattern (like making the water level even everywhere).
Goal B (Safety): Make sure the water never gets so high that it floods the banks (i.e., traffic density never gets so high that it causes a gridlock crash).

2. The Tools: Two Types of "Mathematical Guards"

The authors developed a new way to combine two different mathematical tools to handle both goals at once. They extended these tools from simple systems (like a single car) to complex systems (like a whole highway).

The "Stabilizer" (Control Lyapunov Function - CLF)

Think of this as a GPS Navigator.

What it does: It constantly checks, "Are we where we want to be?" If the traffic density is too high or too low compared to the ideal plan, the GPS tells the speed limit signs to adjust to steer the traffic back to the desired path.
The Limit: A GPS is great at getting you to a destination, but it doesn't always know about a cliff on the side of the road. It might try to drive you straight into a traffic jam just to reach the destination faster.

The "Safety Guard" (Spatial Control Barrier Function - sCBF)

Think of this as a Fence Builder.

What it does: It draws an invisible fence around the "safe zone." It doesn't care about the destination; it only cares that the traffic density never crosses a specific line (the fence). If the traffic gets too close to the fence, this guard immediately slams on the brakes (lowers the speed limit) to push the traffic back to safety.
The Limit: A fence is great for safety, but it doesn't help you get to your destination. It just keeps you from falling off the edge.

3. The Solution: The "Traffic Traffic Cop"

The paper's main innovation is combining the GPS and the Fence into one Traffic Cop using a mathematical "decision-making box" (called Quadratic Programming).

How it works: The Traffic Cop looks at both instructions simultaneously.
- It asks the GPS: "How do we get to the ideal traffic pattern?"
- It asks the Fence: "Are we about to hit the danger zone?"
The Result: The Cop calculates the perfect speed limit adjustment. It tries to follow the GPS as closely as possible, but the moment the GPS tries to drive the traffic into the danger zone, the Fence takes over and overrides the GPS to keep everyone safe.

4. What Happened in the Test?

The researchers tested this on a computer simulation of a 1-kilometer circular road.

Scenario 1 (No Control): Without any help, a traffic jam wave formed and grew, violating safety limits.
Scenario 2 (GPS Only): The system successfully smoothed out the traffic to the desired pattern, but it accidentally drove the density too high, breaking the safety rules.
Scenario 3 (Fence Only): The system kept the traffic safe (never too crowded), but it didn't really fix the traffic flow; it just stayed in a safe, but messy, state.
Scenario 4 (The Combined Cop): This was the winner. The system successfully smoothed the traffic toward the ideal pattern while strictly obeying the safety fence. When the "GPS" tried to push the traffic too hard, the "Fence" gently pulled it back, ensuring the traffic stayed safe without ruining the overall flow.

The Bottom Line

The paper proves that you can mathematically combine "getting to the goal" with "staying safe" for complex, moving systems like traffic. By using these new "spatial" guards, they created a controller that keeps traffic flowing smoothly without ever letting it get dangerously crowded.

Technical Summary: Constrained Control of PDE Traffic Flow via Spatial Control Barrier Functions

Problem Statement
Traffic management systems aim to reduce congestion and regulate traffic density toward a desired profile. However, existing control strategies often fail to explicitly enforce state constraints, such as preventing traffic density from exceeding prescribed safety limits or avoiding extreme velocities. While Variable Speed Limit (VSL) control is a common mechanism for influencing traffic flow in Partial Differential Equation (PDE) models, most approaches either discretize the PDE into Ordinary Differential Equations (ODEs) for controller design or focus solely on regulation without guaranteeing that the system remains within a safe operating range. There is a need for a control framework that operates directly at the PDE level to simultaneously achieve stabilization to a desired profile and enforce safety constraints via forward invariance.

Methodology
The authors develop a unified control framework that extends Control Lyapunov Function (CLF) and Control Barrier Function (CBF) theories from ODE systems to in-domain control of PDEs.

Traffic Model: The study utilizes the Lighthill–Whitham–Richards (LWR) model, a first-order hyperbolic conservation equation describing traffic density ( $\rho$ ) and flow ( $q$ ). The model employs a triangular fundamental diagram. VSL actuation is incorporated by modifying the fundamental diagram with a spatially and temporally varying speed limit ratio, $b(z,t)$ . The control input is defined as the spatial gradient of the VSL ratio, $u = \partial \Delta b / \partial z$ .
Stabilizing Control (CLF): The authors extend CLF theory to PDEs by defining a Lyapunov function $V(t)$ as the quadratic deviation of the traffic density from a desired profile. A feedback control law is derived to ensure the time derivative of $V$ is negative definite, driving the system toward the desired density.
Safe Control (sCBF): To enforce safety, the concept of Control Barrier Functions is adapted for PDEs by introducing Spatially varying Control Barrier Functions (sCBFs). The safe set is defined by a spatially and temporally varying upper bound on density, $\lambda(z,t)$ . An sCBF is constructed to ensure the forward invariance of this set, preventing density from exceeding the prescribed limit.
Unified Optimization Framework: The stabilizing CLF and the safety-enforcing sCBF are unified within a Quadratic Programming (QP) framework. The controller solves an optimization problem at each point in space and time to minimize the deviation from the desired control input while satisfying both the CLF stability condition and the sCBF safety condition. A relaxation variable is included to ensure problem solvability.

Key Contributions

Extension of CLF/CBF to PDEs: The paper extends the theoretical foundations of CLFs and CBFs, previously established for ODEs, to infinite-dimensional systems governed by PDEs.
Spatially Varying Control Barrier Functions: A novel formulation of sCBFs is introduced that allows for safety constraints to be defined and enforced directly at the PDE level, rather than relying on discretized approximations.
Unified Constrained Control: The authors present a QP-based controller that simultaneously achieves regulation to a desired traffic density profile and enforces state constraints (safety) in a single optimization step.
In-Domain VSL Control: The framework is specifically applied to in-domain VSL control of the LWR model, offering a direct method to regulate traffic flow while maintaining safety.

Results
The proposed controllers were validated through a simulation case study on a circular road (length $L=1000$ m) with periodic boundary conditions. Three scenarios were compared:

CLF-only Control: Successfully drove the density toward the desired profile but violated the safety boundary (density limit) when the desired profile or initial conditions pushed the system toward congestion.
sCBF-only Control: Successfully maintained the density within the safe set by lowering speed limits but failed to drive the system to the desired density profile, resulting in significant deviation from the target.
Combined CLF+sCBF Control: This approach successfully balanced both objectives. When the desired profile was within the safe set, the controller acted primarily as a stabilizer. When the desired profile or system dynamics threatened to violate the safety constraint, the sCBF constraint became active, forcing a reduction in speed limits to maintain safety while still driving the system as close as possible to the desired profile.

The results demonstrated that incorporating constrained control minimally affected the stabilizing control input when safety was not at risk, but effectively intervened to maintain density within the defined safe set when necessary.

Significance and Claims
The paper claims to provide a novel formulation for safety-critical constraint enforcement directly at the PDE level, avoiding the potential inaccuracies of discretization-based approaches. By unifying CLF and sCBF theories, the authors offer a method to guarantee that traffic density remains within prescribed limits while simultaneously regulating the system toward a desired state. The work highlights that while barrier functions are well-established for ODEs (e.g., in adaptive cruise control), their application to PDE systems for in-domain control is under-explored. The authors position this work as a step toward robust, safety-guaranteed traffic management, noting that future work will focus on developing sCBFs dependent on control inputs and vehicle speed, and applying these methods to optimal controller designs.