AMPIC: Adaptive Model Predictive Ising Controller

This repository implements AMPIC, a method for controlling traffic signals by solving Ising problems, on the traffic simulator SUMO (Simulation of Urban MObility).

Usage

Run main.py.

python main.py

Option

command	default	description
--nogui	False	run the commandline version of sumo
--input, -i	sq	input file prefix
--path, -p	.	input data path
--controller, -c	4	1:random, 2:pattern, 3:local, 4:global, 5:log
--solver, -v	dwave_sa	strings designating sampler dwave_sa, dwave_qa, dwave_hb, dwave_greedy, amplify_sa, amplify_gurobi, brute_force, or ignore_interaction
--numreads	1000	numreads parameter for d-wave sampler
--tls	traffic_lights_log.csv	input traffic lights data
--threshold, -t	0.1	threshold for local controller
--horizon, -r	1	horizon for global_mpc controller
--step-interval, -s	10	step interval (in unit s) for traffic signal changes
--nored	False	use only green for traffic light
--secs-red	3	length (in unit s) for red traffic light
--secs-yellow	3	length (in unit s) for yellow traffic light
--step-end, -e	3600	total time steps
--weight, -w	0	input weight in hamiltonian
--weight_mode, -q	fixed	state weight in hamiltonian. (fixed / linear / quad)
--freq,	0.5	switching frequency in rondom & pattern controller
--seed	1395	random seed

Citation

If you find our paper or this code useful or relevant to your work please consider citing us.

D. Inoue, H. Yamashita, K. Aihara, and H. Yoshida, “AMPIC: Adaptive Model Predictive Ising Controller for large-scale urban traffic signals.” arXiv, Jun. 05, 2024. doi: 10.48550/arXiv.2406.03690.

In bibtex format:

@misc{ampic,
Author = {Daisuke Inoue and Hiroshi Yamashita and Kazuyuki Aihara and Hiroaki Yoshida},
Title = {AMPIC: Adaptive Model Predictive Ising Controller for large-scale urban traffic signals},
Year = {2024},
Eprint = {arXiv:2406.03690},
}

License

This project is licensed under the GPL2.