Mixed Traffic: A Perspective from Long Duration Autonomy

TL;DR

This paper develops a control strategy for connected autonomous vehicles in mixed traffic at intersections, ensuring safe, timely crossing by integrating optimal control, control barrier functions, and real-time feasibility analysis.

Contribution

It introduces a novel long-duration autonomy controller for CAVs in mixed traffic, combining optimal control, CBFs, and real-time feasibility checks for intersection management.

Findings

The control policy guarantees always feasible solutions.

Simulations validate safety and efficiency in mixed traffic scenarios.

The approach effectively manages intersection crossing times with safety guarantees.

Abstract

The rapid adoption of autonomous vehicle has established mixed traffic environments, comprising both autonomous and human-driven vehicles (HDVs), as essential components of next-generation mobility systems. Along these lines, connectivity between autonomous vehicles and infrastructure (V2I) is also a significant factor that can effectively support higher-level decision-making. At the same time, the integration of V2I within mixed traffic environments remains a timely and challenging problem. In this paper, we present a long-duration autonomy controller for connected and automated vehicles (CAVs) operating in such environments, with a focus on intersections where right turns on red are permitted. We begin by deriving the optimal control policy for CAVs under free-flow traffic. Next, we analyze crossing time constraints imposed by smart traffic lights and map these constraints to controller bounds using Control Barrier Functions (CBFs), with the aim to drive a CAV to cross the intersection on time. We also introduce criteria for identifying, in real-time, feasible crossing intervals for each CAV. To ensure safety for the CAVs, we present model-agnostic safety guarantees, and demonstrate their compatibility with both CAVs and HDVs. Ultimately, the final control actions are enforced through a combination of CBF constraints, constraining CAVs to traverse the intersection within the designated time intervals while respecting other vehicles. Finally, we guarantee that our control policy yields always a feasible solution and validate the proposed approach through extensive simulations in MATLAB.