import random
import numpy as np
import pickle
from env import SoccerEnv
from soccer_stat import SoccerStat
""" This file provides a baseline by using two random agents
"""
# set environment
env = SoccerEnv(width=5, height=5, goal_size=3)
# parameters
EPISODES = 5000
# statistic
stat = SoccerStat()
for i in range(EPISODES):
state = env.reset()
stat.set_initial_ball(state[4])
rewardL = 0
rewardR = 0
done = False
while not done:
# agent 1 decides its action
actionL = random.randint(0, env.act_dim-1)
# agent 2 decides its action
from env import SoccerEnv
from agents.common.training_opponent import StationaryOpponent, RandomSwitchOpponent, RLBasedOpponent
TOP = 0
TOP_RIGHT = 1
RIGHT = 2
BOTTOM_RIGHT = 3
BOTTOM = 4
BOTTOM_LEFT = 5
LEFT = 6
TOP_LEFT = 7
env = SoccerEnv()
agentOP = StationaryOpponent(env_width=env.width, env_height=env.height, env_goal_size=env.goal_size)
state = env.reset()
# loop
env.show()
actionOP = agentOP.get_action(state)
print(actionOP)
done, reward_l, reward_r, state, actions = env.step("type action here!", actionOP)
agentOP.adjust(done, reward_r, i)