try:
env = MicroRTSVecEnv(num_envs=1,
render_theme=2,
ai2s=[microrts_ai.coacAI],
map_path="maps/16x16/basesWorkers16x16.xml",
reward_weight=np.array(
[10.0, 1.0, 1.0, 0.2, 1.0, 4.0]))
# env = gym.make('MicrortsDefeatCoacAIShaped-v3').env
# env = gym.wrappers.RecordEpisodeStatistics(env)
# env.action_space.seed(0)
obs = env.reset()
env.render()
except Exception as e:
e.printStackTrace()
env.action_space.seed(0)
env.reset()
for i in range(10000):
env.render()
action_mask = np.array(env.vec_client.getUnitLocationMasks()).flatten()
time.sleep(0.001)
action = env.action_space.sample()
# optional: selecting only valid units.
if len(action_mask.nonzero()[0]) != 0:
action[0] = action_mask.nonzero()[0][0]
next_obs, reward, done, info = env.step([action])
if done:
env.reset()
env.close()
# ALGO LOGIC: put action logic here
with torch.no_grad():
with sw.timer('inference_value'):
values[step] = agent.get_value(obs[step]).flatten()
with sw.timer('inference_action'):
action, logproba, _, invalid_action_masks[
step] = agent.get_action(obs[step],
envs=envs,
sw=sw)
actions[step] = action.T
logprobs[step] = logproba
with sw.timer('step'):
# TRY NOT TO MODIFY: execute the game and log data.
next_obs, rs, ds, infos = envs.step(action.T)
rewards[step], next_done = rs.view(-1), torch.Tensor(ds).to(
device)
for info in infos:
if 'episode' in info.keys():
print(
f"global_step={global_step}, episode_reward={info['episode']['r']}"
)
writer.add_scalar("charts/episode_reward",
info['episode']['r'], global_step)
# for key in info['microrts_stats']:
# writer.add_scalar(f"charts/episode_reward/{key}", info['microrts_stats'][key], global_step)
break
with sw.timer('train'):
# bootstrap reward if not done. reached the batch limit
render_theme=2,
ai2=microrts_ai.coacAI,
map_path="maps/16x16/basesWorkers16x16.xml",
reward_weight=np.array(
[10.0, 1.0, 1.0, 0.2, 1.0, 4.0, 0.0]))
# env = gym.make('MicrortsDefeatCoacAIShaped-v3').env
# env = gym.wrappers.RecordEpisodeStatistics(env)
# env.action_space.seed(0)
obs = env.reset()
env.render()
except Exception as e:
e.printStackTrace()
# print("unit_locatiuons are at", np.where(env.get_unit_location_mask()==1)[0])
print("reward is", env.step([[17, 2, 0, 3, 0, 1, 2, 123]])[1])
env.render()
# print("unit_locatiuons are at", np.where(env.get_unit_location_mask()==1)[0])
print("reward is", env.step([[34, 4, 1, 2, 1, 2, 3, 109]])[1])
env.render()
# print("unit_locatiuons are at", np.where(env.get_unit_location_mask()==1)[0])
# print("reward is", env.step([12, 0, 0, 0, 0, 0, 0, 0, 0])[1])
# env.render()
# print("unit_locatiuons are at", np.where(env.get_unit_location_mask()==1)[0])
# for i in range(9):
# env.step([12, 0, 0, 0, 0, 0, 0, 0, 0])
# env.render()
# print("unit_locatiuons are at", np.where(env.get_unit_location_mask()==1)[0])
# # # harvest