def test(env):
#TestAgent(env)
total_score = 0
episode_num = 10
max_step = 500
agent = CTRNNAgent()
agent.set_weights(get_initial(0))
counter = 0
total = 100
for i in range(episode_num):
obs = env.reset()
agent.brain.randomize_outputs(0.5,0.51)
for i in range(max_step):
action = agent.Act(obs)
#action = np.array([1,0])
#print("Action:", action)
print("Observation", obs)
if np.isnan(action).any():
return 10
obs, reward, done, info = env.step(action)
if done:
total_score += reward
#print("Done")
#print(reward)
break
return
def test(env):
#TestAgent(env)
total_score = 0
episode_num = 10000
max_step = 500
agent = CTRNNAgent()
agent.set_weights(get_initial(0))
counter = 0
total = 10000
labels = []
for i in range(episode_num):
obs = env.reset()
agent.brain.randomize_outputs(0.5, 0.51)
for i in range(max_step):
action = agent.Act(obs)
counter += 1
if counter % 5 == 0:
#print(counter // 5)
labels.append(obs)
total -= 1
if total == 0:
df = pd.DataFrame(labels)
df.to_csv("./labels.csv", index_label="index")
return labels
#action = np.array([1,0])
#print("Action:", action)
#print("Observation", obs)
if np.isnan(action).any():
return 10
obs, reward, done, info = env.step(action)
if done:
total_score += reward
#print("Done")
#print(reward)
break
return
def EvaluateGenome(genome):
episode_num = 6
max_step = 500
global E
env = E
agent = CTRNNAgent()
agent.set_weights(genome)
total_score = 0
for i in range(episode_num):
obs = env.reset()
agent.brain.randomize_outputs(0.5, 0.51)
for i in range(max_step):
action = agent.Act(obs)
if np.isnan(action).any():
return 10
obs, reward, done, info = env.step(action)
if done:
total_score += reward
break
return -total_score
def test(env):
#TestAgent(env)
total_score = 0
episode_num = 10
max_step = 500
agent = CTRNNAgent()
agent.set_weights(get_initial(0))
print(agent)
for i in range(episode_num):
obs = env.reset()
agent.brain.randomize_outputs(0.5, 0.51)
episode_traj = []
for j in range(max_step):
action, traj = agent.Act(obs, return_trajectory=True)
episode_traj += traj
#action = np.array([1,0])
#print("Action:", action)
#print("Observation", obs)
if np.isnan(action).any():
return 10
obs, reward, done, info = env.step(action)
if done:
traj = np.array(episode_traj)
np.nan_to_num(traj, False)
t = pd.DataFrame(traj)
t.columns = ["I1", "I2", "I3", "M1", "M2"]
t.plot()
plt.show()
total_score += reward
#print("Done")
#print(reward)
break
return