encoderSharpness = 30.0
for v in inputs:
for i in range(0, encoderSize):
center = i / encoderSize * 2.0 - 1.0
delta = center - v
intensity = np.exp(-delta * delta * encoderSharpness)
inputArr.append(intensity)
#reward = dir * paddleX * 0.01
#reward = np.abs(paddleX - ballPosition[0]) < 0.1
a.simStep(reward, 0.002, 0.98, 0.2, 1.0, np.matrix([inputArr]).T, 0.005, 0.005, 0.0005, 0.95, 0.01)
print(a._prevValue)
prevReward = reward
if rewardTimer > 0.0:
rewardTimer -= 1.0
if punishmentTimer > 0.0:
punishmentTimer -= 1.0
paddleX = np.minimum(1.0, np.maximum(0.0, paddleX + 0.15 * np.sum(a.getActions()) * 0.25))
# Render
display.fill((255,255,255))
for v in inputs:
for i in range(0, encoderSize):
center = i / encoderSize * 2.0 - 1.0
delta = center - v
# intensity = np.exp(-delta * delta * encoderSharpness)
intensity = np.absolute(delta) < 0.5 / encoderSize
inputArr.append(intensity)
# reward = dir * paddleX * 0.01
# reward = np.abs(paddleX - ballPosition[0]) < 0.1
a.simStep(reward, 0.001, 0.95, 0.05, np.matrix([inputArr]).T, 0.001, 0.001, 0.01, 0.01, 0.92)
print(a._prevValue)
prevReward = reward
if rewardTimer > 0.0:
rewardTimer -= 1.0
if punishmentTimer > 0.0:
punishmentTimer -= 1.0
paddleX = np.minimum(1.0, np.maximum(0.0, paddleX + 0.2 * np.sum(a.getActions()) / numActions))
# Render
display.fill((255, 255, 255))
for v in inputs:
for i in range(0, encoderSize):
center = i / encoderSize * 2.0 - 1.0
delta = center - v
#intensity = np.exp(-delta * delta * encoderSharpness)
intensity = np.absolute(delta) < 0.5 / encoderSize
inputArr.append(intensity)
#reward = dir * paddleX * 0.01
#reward = np.abs(paddleX - ballPosition[0]) < 0.1
a.simStep(reward, 0.001, 0.95, 0.05,
np.matrix([inputArr]).T, 0.001, 0.001, 0.01, 0.01, 0.92)
print(a._prevValue)
prevReward = reward
if rewardTimer > 0.0:
rewardTimer -= 1.0
if punishmentTimer > 0.0:
punishmentTimer -= 1.0
paddleX = np.minimum(
1.0,
np.maximum(0.0, paddleX + 0.2 * np.sum(a.getActions()) / numActions))
# Render
