def debug_confirm_arena_config(env_path, arena_config):
env_path = (env_path.strip().replace('.app', '').replace(
'.exe', '').replace('.x86_64', '').replace('.x86', ''))
from animalai.envs.environment import UnityEnvironment
env = UnityEnvironment(n_arenas=16,
file_name=env_path,
worker_id=1,
seed=0,
docker_training=False,
play=True)
env.reset(arenas_configurations=arena_config)
try:
while True:
continue
except KeyboardInterrupt:
env.close()
def main(args):
docker_training = docker_target_name is not None
env = UnityEnvironment(
n_arenas=args.n_arenas,
file_name=env_path,
worker_id=worker_id,
seed=seed,
docker_training=docker_training,
play=False,
resolution=resolution
)
arena_config_in = ArenaConfig('configs/3-Obstacles.yaml')
env.reset(arenas_configurations=arena_config_in)
start_time = time.time()
for i in range(args.frames):
res = env.step(np.random.randint(0, 3, size=2 * args.n_arenas))
elapsed_time = time.time() - start_time
fps = float(args.frames) / elapsed_time
print("n_arenas={0}, fps={1:.3f}".format(args.n_arenas, fps))
env.close()
# let the agent generate an action based on the information
action = agent.step(obs, reward, done, info)
# Visualization{visual, direction, path}
image.set_data(obs[0])
if agent.chaser.newest_path is not None:
sca.set_offsets(np.array(agent.chaser.newest_path))
else:
sca.set_offsets(AgentConstants.standpoint[::-1])
if agent.chaser.newest_end is not None:
line.set_xdata(
[AgentConstants.standpoint[1], agent.chaser.newest_end[0]])
line.set_ydata(
[AgentConstants.standpoint[0], agent.chaser.newest_end[1]])
else:
line.set_xdata([])
line.set_ydata([])
fig.canvas.draw()
fig.canvas.flush_events()
# go to next test if the current one is finised
if all(brainInfo['Learner'].local_done):
break
else:
brainInfo = env.step(action)
# cleanup
plt.close(fig)
env.close()