traci.trafficlight.Phase(32, "rrGrrrrrGrrr"),
traci.trafficlight.Phase(2, "rryrrrrryrrr"),
traci.trafficlight.Phase(32, "rrrGGrrrrGGr"),
traci.trafficlight.Phase(2, "rrryyrrrryyr"),
traci.trafficlight.Phase(32, "rrrrrGrrrrrG"),
traci.trafficlight.Phase(2, "rrrrryrrrrry")
])
if args.reward == 'queue':
env._compute_rewards = env._queue_average_reward
else:
env._compute_rewards = env._waiting_time_reward
for run in range(1, args.runs + 1):
initial_states = env.reset()
ql_agents = {
ts: QLAgent(starting_state=env.encode(initial_states[ts]),
state_space=env.observation_space,
action_space=env.action_space,
alpha=args.alpha,
gamma=args.gamma,
exploration_strategy=EpsilonGreedy(
initial_epsilon=args.epsilon,
min_epsilon=args.min_epsilon,
decay=args.decay))
for ts in env.ts_ids
}
done = {'__all__': False}
infos = []
if args.fixed:
while not done['__all__']:
max_green=args.max_green,
max_depart_delay=0,
phases=[
traci.trafficlight.Phase(ns, "GGrr"), # north-south
traci.trafficlight.Phase(2000, "yyrr"),
traci.trafficlight.Phase(we, "rrGG"), # west-east
traci.trafficlight.Phase(2000, "rryy")
])
if args.reward == 'queue':
env._compute_rewards = env._queue_average_reward
else:
env._compute_rewards = env._waiting_time_reward
for run in range(1, args.runs+1):
initial_states = env.reset()
ql_agents = {ts: QLAgent(starting_state=env.encode(initial_states[ts]),
state_space=env.observation_space,
action_space=env.action_space,
alpha=args.alpha,
gamma=args.gamma,
exploration_strategy=EpsilonGreedy(initial_epsilon=args.epsilon, min_epsilon=args.min_epsilon, decay=args.decay)) for ts in env.ts_ids}
done = {'__all__': False}
infos = []
if args.fixed:
while not done['__all__']:
_, _, done, _ = env.step({})
else:
while not done['__all__']:
actions = {ts: ql_agents[ts].act() for ts in ql_agents.keys()}
max_depart_delay=0,
phases=[
traci.trafficlight.Phase(args.ns, "GGrr"), # north-south
traci.trafficlight.Phase(2, "yyrr"),
traci.trafficlight.Phase(args.we, "rrGG"), # west-east
traci.trafficlight.Phase(2, "rryy")
])
if args.reward == 'queue':
env._compute_rewards = env._queue_average_reward
else:
env._compute_rewards = env._waiting_time_reward
for run in range(1, args.runs + 1):
initial_states = env.reset()
ql_agents = {
ts: QLAgent(starting_state=env.encode(initial_states[ts]),
state_space=env.observation_space,
action_space=env.action_space,
alpha=args.alpha,
gamma=args.gamma,
exploration_strategy=EpsilonGreedy(
initial_epsilon=args.epsilon,
min_epsilon=args.min_epsilon,
decay=args.decay))
for ts in env.ts_ids
}
done = {'__all__': False}
infos = []
if args.fixed:
while not done['__all__']:
traci.trafficlight.Phase(args.ns, "GGrr"), # north-south
traci.trafficlight.Phase(2, "yyrr"),
traci.trafficlight.Phase(args.we, "rrGG"), # west-east
traci.trafficlight.Phase(2, "rryy")
])
env = VisualizationEnv(
env=env,
episodic=False,
features_names=['Phase 0', 'Phase 1', 'Elapsed time'] + ['Density lane ' + str(i) for i in range(4)] + ['Queue lane ' + str(i) for i in range(4)],
actions_names=['Phase 0', 'Phase 1']
)
for run in range(1, args.runs+1):
initial_states = env.reset()
ql_agents = {ts: QLAgent(starting_state=env.encode(initial_states),
state_space=env.observation_space,
action_space=env.action_space,
alpha=args.alpha,
gamma=args.gamma,
exploration_strategy=EpsilonGreedy(initial_epsilon=args.epsilon, min_epsilon=args.min_epsilon, decay=args.decay)) for ts in env.ts_ids}
env.set_agent(ql_agents['t'])
env.add_plot('Epsilon', lambda: ql_agents['t'].exploration.epsilon)
done = False
while not done:
actions = {ts: ql_agents[ts].act() for ts in ql_agents.keys()}
s, r, done, _ = env.step(action=actions['t'])
for agent_id in ql_agents.keys():
ql_agents[agent_id].learn(next_state=env.encode(s), reward=r)