Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Traffic signal optimization in large-scale urban road networks: an adaptive-predictive controller using Ising models

Daisuke Inoue, Hiroshi Yamashita, Kazuyuki Aihara, Hiroaki Yoshida

TL;DR

This work tackles the challenge of scalable adaptive traffic-signal control in large urban networks. It proposes AMPIC, an Adaptive Model Predictive Ising Controller that transforms a multi-horizon MPC objective into a binary Ising problem with $N k_h$ variables and solves it using Ising solvers, including quantum annealing. The approach leverages an adaptive internal model of vehicle flow and demonstrates, via SUMO simulations on realistic Sapporo-like networks and lattice topologies, that AMPIC improves mean vehicle speed, reduces waiting times, and lowers CO2 emissions, with the benefits growing for larger networks and longer prediction horizons. The results indicate that Ising-based MPC can enable globally optimized, decarbonization-friendly traffic management at city scale, provided sensing and computational resources are available, and point to future extensions with connected/autonomous vehicles and broader solver comparisons.

Abstract

Realizing smooth traffic flow is important for achieving carbon neutrality. Adaptive traffic signal control, which considers traffic conditions, has thus attracted attention. However, it is difficult to ensure optimal vehicle flow throughout a large city using existing control methods because of their heavy computational load. Here, we propose a control method called AMPIC (Adaptive Model Predictive Ising Controller) that guarantees both scalability and optimality. The proposed method employs model predictive control to solve an optimal control problem at each control interval with explicit consideration of a predictive model of vehicle flow. This optimal control problem is transformed into a combinatorial optimization problem with binary variables that is equivalent to the so-called Ising problem. This transformation allows us to use an Ising solver, which has been widely studied and is expected to have fast and efficient optimization performance. We performed numerical experiments using a microscopic traffic simulator for a realistic city road network. The results show that AMPIC enables faster vehicle cruising speed with less waiting time than that achieved by classical control methods, resulting in lower CO2 emissions. The model predictive approach with a long prediction horizon thus effectively improves control performance. Systematic parametric studies on model cities indicate that the proposed method realizes smoother traffic flows for large city road networks. Among Ising solvers, D-Wave's quantum annealing is shown to find near-optimal solutions at a reasonable computational cost.

Traffic signal optimization in large-scale urban road networks: an adaptive-predictive controller using Ising models

TL;DR

This work tackles the challenge of scalable adaptive traffic-signal control in large urban networks. It proposes AMPIC, an Adaptive Model Predictive Ising Controller that transforms a multi-horizon MPC objective into a binary Ising problem with variables and solves it using Ising solvers, including quantum annealing. The approach leverages an adaptive internal model of vehicle flow and demonstrates, via SUMO simulations on realistic Sapporo-like networks and lattice topologies, that AMPIC improves mean vehicle speed, reduces waiting times, and lowers CO2 emissions, with the benefits growing for larger networks and longer prediction horizons. The results indicate that Ising-based MPC can enable globally optimized, decarbonization-friendly traffic management at city scale, provided sensing and computational resources are available, and point to future extensions with connected/autonomous vehicles and broader solver comparisons.

Abstract

Realizing smooth traffic flow is important for achieving carbon neutrality. Adaptive traffic signal control, which considers traffic conditions, has thus attracted attention. However, it is difficult to ensure optimal vehicle flow throughout a large city using existing control methods because of their heavy computational load. Here, we propose a control method called AMPIC (Adaptive Model Predictive Ising Controller) that guarantees both scalability and optimality. The proposed method employs model predictive control to solve an optimal control problem at each control interval with explicit consideration of a predictive model of vehicle flow. This optimal control problem is transformed into a combinatorial optimization problem with binary variables that is equivalent to the so-called Ising problem. This transformation allows us to use an Ising solver, which has been widely studied and is expected to have fast and efficient optimization performance. We performed numerical experiments using a microscopic traffic simulator for a realistic city road network. The results show that AMPIC enables faster vehicle cruising speed with less waiting time than that achieved by classical control methods, resulting in lower CO2 emissions. The model predictive approach with a long prediction horizon thus effectively improves control performance. Systematic parametric studies on model cities indicate that the proposed method realizes smoother traffic flows for large city road networks. Among Ising solvers, D-Wave's quantum annealing is shown to find near-optimal solutions at a reasonable computational cost.
Paper Structure (11 sections, 21 equations, 8 figures, 1 table, 1 algorithm)

This paper contains 11 sections, 21 equations, 8 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Method
  3. Model predictive control of traffic signals
  4. Ising models, problems, and solvers
  5. Coupled simulation with SUMO
  6. Results
  7. Evaluation criteria
  8. Results for Sapporo
  9. Size of road network
  10. Prediction horizon impact
  11. Discussion

Figures (8)

  • Figure 1: Schematic diagram of the proposed method (AMPIC) and snapshots of vehicles in SUMO with various traffic control methods. (a) Vehicle information collected from the traffic system is sent to AMPIC and control quantities computed by AMPIC are sent back (see Section \ref{['sec:method-true']} for explanation of variables). (b) Snapshots of vehicles in numerical evaluation with SUMO on roads that replicate urban areas in Sapporo, Hokkaido, Japan. The vehicle generation rate was set to $2.22$ vehicles per second. The snapshots represent the state of vehicles one hour after the start of the simulation. The color of a vehicle represents its speed. The results for traditional pattern control, local switching control, and control with AMPIC are shown. The total CO2 emissions for each control method are also shown.
  • Figure 2: Definition of variables for the road network.
  • Figure 3: Road network created to replicate an urban area of Sapporo, Hokkaido, Japan. The base map image is provided by © OpenStreetMap OSMOSMPaperOSMap.
  • Figure 4: Time evolution of performance indicators for various methods. The proposed method (blue solid) is compared with local control (orange dash-dotted), random control (green dotted), and pattern control (red dashed). The colored area represents the standard error of each performance indicator for various seeds of the random number used for vehicle route generation.
  • Figure 5: Time-averaged performance indicators for various vehicle generation rates for various methods. The proposed method (blue solid) is compared with local control (orange dash-dotted), random control (green dotted), and pattern control (red dashed). The error bars represent the standard error of each performance indicator for various seeds of the random number used for vehicle route generation.
  • ...and 3 more figures