Towards Distributed Mixed-Integer Quadratic Programming for Mixed-Traffic Intersection Control with Proximal ADMM

Authors: Viet-Anh Le and Andreas A. Malikopoulos

Summary

In this paper, we propose a distributed algorithm based on the proximal alternating direction method of multipliers (ADMM) and sequential constraint tightening to solve mixed-integer quadratic programming (MIQP) problems arising from traffic light systems and connected automated vehicles (CAVs) in mixed-traffic intersections. We formulate an MIQP problem to jointly coordinate traffic light systems and CAVs in mixed-traffic intersections, fully leveraging the benefits of CAV coordination under high penetration rates. To approximately solve the complex multi-agent MIQP problem, we develop a distributed algorithm that applies proximal ADMM to solve the convex relaxation of the MIQP and sequentially tightens the constraint coefficients to enforce integrality constraints. We validate the performance of the control framework and the distributed algorithm through simulations across different penetration rates and traffic volumes.