Intervention Strategies for Fairness and Efficiency at Autonomous Single-Intersection Traffic Flows

Salman Ghori, Ania Adil, Melkior Ornik, Eric Feron · 2025

Read Intervention Strategies for Fairness and Efficiency at Autonomous Single-Intersection Traffic Flows by Salman Ghori, Ania Adil, Melkior Ornik, Eric Feron (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-02562.txt.

Introduction

Salman Ghori and three co-authors study self driving cars crossing a junction with no traffic lights. Junctions are a hard point in traffic control, where safety and smooth flow both matter. Gains in speed often come at the cost of fairness, which users need to trust the system.

Their 2025 study asks when a central controller should step in to guide autonomous agents. Timing that intervention changes safety, efficiency, and fairness at once, under strict safety limits. The findings speak to city planners who must balance fast traffic with equal treatment for every driver. The right time to step in may shift with load, speed, and the mix of cars on the road that day.

Catalog record

CREATORSalman Ghori, Ania Adil, Melkior Ornik, Eric Feron
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-02562

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Intervention Strategies for Fairness and Efficiency at Autonomous Single-Intersection Traffic Flows Salman Ghori, Ania Adil, Melkior Ornik, Eric Feron Abstract Intersections present significant challenges in traffic management, where ensuring safety and efficiency is essential for effective flow. However, these goals are often achieved at the expense of fairness, which is critical for trustworthiness and long-term sustainability. This paper investigates how the timing of centralized intervention affects the management of autonomous agents at a signal-less, orthogonal intersection, while satisfying safety constraints, evaluating efficiency, and ensuring fairness. A mixed-integer linear programming (MILP) approach is used to optimize agent coordination within a circular control zone centered at the intersection. We introduce the concept of fairness, measured via pairwise reversal counts, and incorporate fairness constraints into the MILP framework. We then study the relationship between fairness and system efficiency and its impact on platoon formation. Finally, simulation studies analyze the effectiveness of early versus late intervention strategies and fairness-aware control, focusing on safe, efficient, and robust management of agents within the control zone. Nomenclature Indices, Sets, and Spaces \mathbb{R} Set of real numbers. \mathbb{R}_{+} Nonnegative reals: \{\,x\in\mathbb{R}:x\geq 0\} . \mathbb{R}_{+}^{\ast} Strictly positive reals: \{\,x\in\mathbb{R}:x>0\} . \mathbb{R}^{n} n -dimensional Euclidean space (column vectors). \mathbb{R}^{n}_{+} Nonnegative orthant: \{\,x\in\mathbb{R}^{n}:x_{i}\geq 0\;\forall\,i\} . \mathbb{R}^{n\times m} Set of real n\times m matrices. G Total number of agents. N Total number of discrete time steps in the horizon. q,r Agent indices, q,r\in\{\,1,\dots,G\} . i Discrete time index, i\in\{\,0,\dots,N\} . Decision Variables p_{q,i}^{x},\,p_{q,i}^{y}\in\mathbb{R} Position [m] of agent q in the x - and y -coordinates at time i . v_{q,i}^{x},\,v_{q,i}^{y}\in\mathbb{R} Velocity [m/s] of agent q in the x - and y -directions at time i . x_{q,i}\in\mathbb{R}^{4} State vector of agent q at time i , x_{q,i}=\bigl(p_{q,i}^{x},\;p_{q,i}^{y},\;v_{q,i}^{x},\;v_{q,i}^{y}\bigr)^{\top}. u_{q,i}\in\mathbb{R}^{2} Control input [m/s 2 ] of agent q at time i , u_{q,i}=\bigl(u_{q,i}^{x},\;u_{q,i}^{y}\bigr)^{\top}. s_{q,i,k}\in\mathbb{R}_{+} , k=1,\dots,4 Slack variables for the absolute deviation of each state component from its desired final value at time i . V_{q,i}\in\mathbb{R}_{+} Velocity [m/s] of agent q at time i ; used in the objective to reward higher speed. O_{q,r}\in\{0,1\} Binary variable indicating priority reversal between agents q and r . \varpi_{q,i},\,\varpi_{r,i},\,\mathrm{cl}_{q,r,i}\in\{0,1\} Binary variables d_{q,i},\,d_{r,i}\,\delta_{q,r,i}\in\mathbb{R}_{+} Auxiliary variables Parameters and Constants R\in\mathbb{R}_{+} Radius [m] of the control zone. x_{q,0}\in\mathbb{R}^{4} Initial state of agent q : x_{q,0}=\bigl(p_{q,0}^{x},\;p_{q,0}^{y},\;v_{q,0}^{x},\;v_{q,0}^{y}\bigr)^{\top}. x_{q,f}\in\mathbb{R}^{4} Desired final state of agent q : x_{q,f}=\bigl(p_{q,f}^{x},\;p_{q,f}^{y},\;v_{q,f}^{x},\;v_{q,f}^{y}\bigr)^{\top}. A_{q}\in\mathbb{R}^{4\times 4},\,B_{q}\in\mathbb{R}^{4\times 2} State‐transition and control‐input matrices for agent q . u_{q,\min},\,u_{q,\max}\in\mathbb{R}^{2} Lower and upper bounds on control inputs for agent q x_{q,\min},\,x_{q,\max}\in\mathbb{R}^{4} Lower and upper bounds on the state variables of agent q \alpha,\,\beta\in\mathbb{R}_{+}^{4} Weight vectors for state deviation. \gamma\in\mathbb{R}_{+} Weight for velocity V_{q,i} . \lambda\in\mathbb{R}_{+} Weight for priority‐reversal O_{q,r} . c_{x},c_{y}\in\mathbb{R} Coordinates of the intersection center. d_{\mathrm{safe}}\in\mathbb{R}_{+}^{\ast} Minimum safety distance [m] between any two agents within the same flow. d_{\mathrm{sep}}\in\mathbb{R}_{+}^{\ast} Minimum cross-flow separation distance [m] at the intersection. \Lambda_{x} , \Lambda_{y} Poisson arrival rates [s -1 ] for x–flow, y–flow. b\in\mathbb{R}_{+} Threshold to identify possible reversal of agents q,r . W,M\in\mathbb{R}_{+} Sufficiently large constant (“Big‐ M ”). \epsilon\in\mathbb{R}_{+}^{\ast} Small positive constant. d_{\mathrm{stop}}\in\mathbb{R}_{+}^{\ast} Safe braking distance [m] to stop completely (worst-case). \delta_{\mathrm{platoon}}\in\mathbb{R}_{+} Length of the longest platoon. n_{\mathrm{platoon}}\in\mathbb{R}_{+} Largest trailing agent count in either flow. n_{1},n_{2}\in\mathbb{R}_{+} Platoon sizes in x/y–flows. I Introduction Intersections represent constrained spaces where two or more distinct flows converge, presenting significant challenges and opportunities in traffic management [ 1 , 2 , 3 , 4 , 5 , 6 ] . Effective traffic management is important to maximize capacity and throughput, reduce overall delay, ensure safety, and improve fuel efficiency. Researchers have extensively investigated both light-controlled [ 7 , 8 , 9 , 10 ] and signal-less traffic intersections [ 11 , 12 , 13 , 14 , 15 ] . However, as transportation evolves towards greater autonomy, managing intersections without traffic signals becomes increasingly applicable across diverse contexts, including autonomous vehicle fleets at intersecting streets [ 16 , 17 ] , satellites or spacecraft in different orbital planes that may have the same crossing windows [ 18 , 19 ] , airport runway scheduling [ 20 , 21 , 22 , 23 , 24 ] , and automated warehouse robot sorting goods [ 25 ] . Despite differences in scale and domain, the above-mentioned scenarios share a common structural challenge in coordinating intersecting flows to prevent collisions while optimizing efficiency. A canonical example involves two orthogonal flows crossing a signal-less intersection, which raises key questions about safety guarantees and performance objectives. In particular, the timing of intervention by a centralized controller to manage the intersection has received little attention in the literature. Related work that has a similar theme to the timing of intervention includes Tlig et al.’s synchronization‐based intersection controller [ 26 ] and Lim et al.’s Extended Arrival Manager for air traffic [ 27 ] , although not an intersection controller, shares the key idea of earlier centralized coordination. Beyond efficiency, effective traffic management must address fairness, essential for operational trustworthiness [ 28 ] , stakeholder endorsement [ 29 ] , and long-term sustainability [ 30 ] . Ignoring fairness risks systemic discrimination and stakeholder discontent [ 31 , 32 ] , particularly under high traffic density conditions. Although intuitive, fairness usually lacks a universal formal definition [ 33 ] because perceptions of fairness can vary based on context, perspective, or objectives. Fairness metrics [ 34 ] can be categorized as reversal [ 35 ] , overtaking [ 36 ] , and time-ordered deviation [ 37 ] . Importantly, when agents cross the intersection in an order different from their arrival order at the control zone, not strictly following the first-in–first-out (FIFO) rule, they undergo a reversal. This can facilitate platooning, i.e., short-headway groups of agents with near-synchronized velocities [ 38 , 39 ] . These dense formations cross as units, improving throughput [ 40 ] , reducing delay [ 41 ] , and lowering energy consumption [ 42 ] . In our prior work [ 43 ] , we studied early versus late intervention, modeled as large versus small control zones, at a fixed traffic density and identified a delay-minimizing optimal zone radius. We empirically showed that performance saturates beyond a threshold radius. We also observed platoon formation driven by reversal-enabled sequencing under collision constraints. While our previous study successfully demonstrated these efficiency gains, it did not evaluate the consequential impacts on system fairness. Building on these findings, the present work develops a systematic study of fairness across intervention strategies using MILP. While early intervention offers greater scheduling flexibility that may improve fairness, late intervention can better respect FIFO order, but often at an efficiency cost. We address this gap by elevating fairness to a core objective and examining how fairness constraints interact with control zone sizing across traffic densities. Within our framework, collision-avoidance constraints naturally encourage platooning. Allowing reversals near the intersection can delay one agent slightly to let another pass first, releasing a tightly packed sequence behind it and compressing flows for efficiency gains. Conversely, prioritization of FIFO suppresses platooning, improving fairness but reducing efficiency. This paper significantly expands our framework with three key contributions: • We explicitly integrate fairness constraints into the intersection management framework, demonstrating how it reshapes scheduling decisions and reveals equity–efficiency trade-offs. • We show how control zone size critically influences

Read the complete text