def __init__(self, name, config):
env.__init__(self, name)
self.env = gym.make(name)
self.history_step = config.history_step
self.ob_dims = config.ob_dims
self.act_repeat = config.act_repeat
self.history = []
def __init__(self, name, config):
env.__init__(self, name)
#self.env = gym.make(name)
self.inner_step = 0
self.config = config
# load data
self.data_loader = DataLoader(
self.config.task_data_path,self.config.trajectory_data_path)
self.data_loader.load_task_static() # 读取yellow 数据
self.data_loader.get_trajectories() # 读取uber数据
#self.data_loader.overall_position_normalization()
self.data_loader.get_merge_task(self.config.aim_day_num)
print "task generation"
self.task_generator = TaskGenerator()
self.task_generator.gen_task_list(self.data_loader.zip_data) # 按分布采样生成task列表
self.task_generator.set_poisson_distribution(self.config.poisson_lamda, self.config.poisson_episode_num)
print "trajectory sampling size: %d" % len(self.data_loader.trajectory_data)
# new simulator
self.simulator = StateSimulator()
# reset
self.simulator.trajector.init_sampling( \
self.data_loader.trajectory_data, self.config.trajector_sampling_size) # 采样生成路线数据
self.episode_task_num = self.config.episode_task_num
self.speed_init()
#self.task_sampling()
#self.preprocess()
# clear memeory
# self.self.data_loader.reset()
# clear log
self.log_file = codecs.open(self.config.log_file_path, "w", "utf8")
self.log_file.close()
def __init__(self, env_config):
env.__init__(self, "trajEnv")
#self.config = config
self.config = self.load_config(env_config['config_path'])
#self.env = gym.make(name)
self.action_space = spaces.MultiDiscrete([self.config['max_par_size']] * self.config['max_task_size'])
# #print self.action_space
# par_feature = np.zeros((self.config['max_par_size'], self.config['par_feature_size']))
# task_feature = np.zeros((self.config['max_task_size'], self.config['task_feature_size']))
obs_size = self.config['max_par_size'] + self.config['max_task_size']
feature_size = max(self.config['task_feature_size'], self.config['par_feature_size'])
#self.observation_space = Box(low=-200, high=200, shape=(obs_size, feature_size), dtype=np.float64)
self.observation_space = spaces.Tuple([
# spaces.Box(
# low=-2000, high=2000, shape=(self.config['max_par_size'], self.config['par_feature_size']), dtype=np.float64),
# spaces.Box(
# low=-2000, high=2000, shape=(self.config['max_task_size'], self.config['task_feature_size']), dtype=np.float64),
spaces.Box(
low=-20, high=20, shape=(self.config['max_task_size'], self.config['max_par_size']), dtype=np.float64),
spaces.Box(
low=-20, high=20, shape=(self.config['max_task_size'], self.config['max_par_size']), dtype=np.float64)
])
#print self.observation_space
self.inner_step = 0
# load data
self.data_loader = DataLoader(
self.config['task_data_path'],self.config['trajectory_data_path'])
self.data_loader.load_task_static() # 读取yellow 数据
self.data_loader.get_trajectories() # 读取uber数据
self.data_loader.overall_position_normalization()
self.data_loader.get_merge_task(self.config['aim_day_num'])
print "task generation"
self.task_generator = TaskGenerator()
self.task_generator.gen_task_list(self.data_loader.zip_data) # 按分布采样生成task列表
self.task_generator.set_poisson_distribution(self.config['poisson_lamda'], self.config['poisson_episode_num'])
print "trajectory sampling size: %d" % len(self.data_loader.trajectory_data)
# new simulator
self.simulator = StateSimulator()
# reset
self.simulator.trajector.init_sampling( \
self.data_loader.trajectory_data, self.config['trajector_sampling_size']) # 采样生成路线数据
self.episode_task_num = self.config['episode_task_num']
self.speed_init()
#self.task_sampling()
#self.preprocess()
# clear memeory
# self.self.data_loader.reset()
# clear log
self.log_file = codecs.open(self.config['log_file_path'], "w", "utf8")
self.log_file.close()
def __init__(self, name, config):
env.__init__(self, name)
#self.env = gym.make(name)
self.inner_step = 0
self.config = config
# load data
self.data_loader = DataLoader(self.config.task_data_path,
self.config.trajectory_data_path)
self.data_loader.load_task_static() # 读取yellow 数据
self.data_loader.get_trajectories() # 读取uber数据
self.data_loader.overall_position_normalization()
self.data_loader.get_merge_task(self.config.aim_day_num)
print "task generation"
self.task_generator = TaskGenerator()
self.task_generator.gen_task_list(
self.data_loader.zip_data) # 按分布采样生成task列表
self.task_generator.set_poisson_distribution(
self.config.poisson_lamda, self.config.poisson_episode_num)
print "trajectory sampling size: %d" % len(
self.data_loader.trajectory_data)
# energy
self.energy = [
3, 2, 2, 5, 1, 7, 5, 8, 9, 4, 6, 10, 6, 4, 7, 3, 1, 8, 4, 6, 10, 3,
6, 4, 3, 6, 2, 2, 10, 1, 1, 1, 2, 4, 9, 6, 5, 3, 4, 7, 8, 2, 6, 1,
8, 6, 2, 9, 10, 9, 5, 2, 4, 1, 9, 1, 1, 3, 1, 4, 4, 8, 9, 9, 5, 8,
10, 6, 2, 8, 4, 2, 8, 4, 3, 1, 10, 2, 1, 5, 6, 6, 10, 2, 9, 6, 2,
5, 4, 10, 4, 1, 9, 9, 6, 5, 10, 7, 10, 7
]
# for i in range(self.config.participant_num):
# energy_save = random.randint(1, 10)
# self.energy.append(energy_save)
# new simulator
self.simulator = StateSimulator()
# reset
self.simulator.trajector.init_sampling( \
self.data_loader.trajectory_data, self.config.trajector_sampling_size) # 采样生成路线数据
self.episode_task_num = self.config.episode_task_num
self.speed_init()
#self.task_sampling()
#self.preprocess()
# clear memeory
# self.self.data_loader.reset()
# clear log
self.log_file = codecs.open(self.config.log_file_path, "w", "utf8")
self.log_file.close()
def __init__(self, env_config):
env.__init__(self, "trajEnv")
#self.config = config
self.config = self.load_config(env_config['config_path'])
self.config["max_step"] = env_config["max_step"]
self.config["episode_task_num"] = env_config["episode_task_num"]
self.config["max_task_size"] = env_config["max_task_size"]
self.config["participant_num"] = env_config["participant_num"]
self.config["max_par_size"] = env_config["max_par_size"]
#self.env = gym.make(name)
self.action_space = spaces.MultiDiscrete([self.config['max_par_size']] * self.config['max_task_size'])
# #print self.action_space
# par_feature = np.zeros((self.config['max_par_size'], self.config['par_feature_size']))
# task_feature = np.zeros((self.config['max_task_size'], self.config['task_feature_size']))
obs_size = self.config['max_par_size'] + self.config['max_task_size']
feature_size = max(self.config['task_feature_size'], self.config['par_feature_size'])
#self.observation_space = Box(low=-200, high=200, shape=(obs_size, feature_size), dtype=np.float64)
self.observation_space = spaces.Tuple([
# spaces.Box(
# low=-2000, high=2000, shape=(self.config['max_par_size'], self.config['par_feature_size']), dtype=np.float64),
# spaces.Box(
# low=-2000, high=2000, shape=(self.config['max_task_size'], self.config['task_feature_size']), dtype=np.float64),
spaces.Box(
low=-20, high=20, shape=(self.config['max_task_size'], self.config['max_par_size']), dtype=np.float64),
spaces.Box(
low=-20, high=20, shape=(self.config['max_task_size'], self.config['max_par_size']), dtype=np.float64)
])
#print self.observation_space
self.inner_step = 0
# load data
self.data_loader = DataLoader(
self.config['task_data_path'],self.config['trajectory_data_path'])
self.data_loader.load_task_static() # 读取yellow 数据
self.data_loader.get_trajectories() # 读取uber数据
self.data_loader.overall_position_normalization()
self.data_loader.get_merge_task(self.config['aim_day_num'])
print "task generation"
self.task_generator = TaskGenerator()
self.task_generator.gen_task_list(self.data_loader.zip_data) # 按分布采样生成task列表
self.task_generator.set_poisson_distribution(self.config['poisson_lamda'], self.config['poisson_episode_num'])
print "trajectory sampling size: %d" % len(self.data_loader.trajectory_data)
# energy
self.energy = [3, 2, 2, 5, 1, 7, 5, 8, 9, 4, 6, 10, 6, 4, 7, 3, 1, 8, 4, 6, 10, 3, 6, 4, 3, 6, 2, 2, 10, 1, 1, 1, 2, 4, 9, 6, 5, 3, 4, 7, 8, 2, 6, 1, 8, 6, 2, 9, 10, 9, 5, 2, 4, 1, 9, 1, 1, 3, 1, 4, 4, 8, 9, 9, 5, 8, 10, 6, 2, 8, 4, 2, 8, 4, 3, 1, 10, 2, 1, 5, 6, 6, 10, 2, 9, 6, 2, 5, 4, 10, 4, 1, 9, 9, 6, 5, 10, 7, 10, 7]
# for i in range(self.config.participant_num):
# energy_save = random.randint(1, 10)
# self.energy.append(energy_save)
# clear memeory
# new simulator
self.simulator = StateSimulator()
# reset
self.simulator.trajector.init_sampling( \
self.data_loader.trajectory_data, self.config['trajector_sampling_size']) # 采样生成路线数据
self.episode_task_num = self.config['episode_task_num']
self.speed_init()
#self.task_sampling()
#self.preprocess()
print "parm## %d %d %d %d %d " % (self.config["max_step"], \
self.config["episode_task_num"], self.config["max_task_size"], \
self.config["participant_num"], self.config["max_par_size"])
# self.self.data_loader.reset()
# clear log
self.log_file = codecs.open(self.config['log_file_path'], "w", "utf8")
self.log_file.close()
