Optimize waiting time of cross using Q-Learning.
- install dependencies.
pip3 install pywin32
pip3 install numpy- run whole program.
python main.py- use class Cross in env.py
from env import Cross
#initializing instance
cross = Cross(inp_file, ini_file)
#initializing simulator
cross.simulation_config(40, 0, 3600)
cross.reset()
#set vehicle input
vehicle_inputs = np.array([244, 198, 268, 216, 132, 210, 272, 152])
cross.set_vehicle_input(vehicle_inputs)
#simulate a cycletime
[status_, reward] = cross.env_interface(70, ql.action_table[action])- use class QL in q_learn.py
from q_learn import QL
#initializing instance
ql = QL()
#load exsitent q and p table.
ql.load_data()
#learn
ql.learn(status, action, reward, status_)
#save q and p table
ql.save_data()
#make choice using the trained q table.
action = ql.check(status)
MIT © Hoream