Skip to content

Latest commit

 

History

History
45 lines (36 loc) · 1.03 KB

File metadata and controls

45 lines (36 loc) · 1.03 KB

Multi-Commander

Multi-agent signal control

Implementation of DQN, Double DQN and Dueling DQN

usage

Training

DQN

$ python run_rl_control.py --algo DQN --epoch 200 --num_step 1500 --phase_step 1

Double DQN

$ python run_rl_control.py --algo DDQN --epoch 100 --num_step 1500 --phase_step 1

Dueling DQN

$ python run_rl_control.py --algo DuelDQN --epoch 10

Inference

DQN

$ python run_rl_control.py --algo DQN --inference --num_step 2000 --ckpt model/DQN_20190731_144939/DQN-200.h5
$ python run_rl_control.py --algo DQN --inference --num_step 2000 --ckpt model/DQN_20190801_124826/DQN-100.h5

DDQN

$ python run_rl_control.py --algo DDQN --inference --num_step 2000 --ckpt model/DDQN_20190801_085209/DDQN-100.h5

Dueling DQN

$ python run_rl_control.py --algo DuelDQN --inference --num_step 2000 --ckpt model/DuelDQN_20190730_165409/DuelDQN-ckpt-10

Simulation

$ . simulation.sh
open firefox with the url: http://localhost:8080/?roadnetFile=roadnet.json&logFile=replay.txt