def composeFowardOneTimeStepWithRandomSubtlety(numOfAgent, idx):
# experiment parameter for env
numMDPTimeStepPerSecond = 5 # change direction every 200ms
distanceToVisualDegreeRatio = 20
minSheepSpeed = int(17.4 * distanceToVisualDegreeRatio / numMDPTimeStepPerSecond)
maxSheepSpeed = int(23.2 * distanceToVisualDegreeRatio / numMDPTimeStepPerSecond)
warmUpTimeSteps = 10 * numMDPTimeStepPerSecond # 10s to warm up
prepareSheepVelocity = PrepareSheepVelocity(minSheepSpeed, maxSheepSpeed, warmUpTimeSteps)
minWolfSpeed = int(8.7 * distanceToVisualDegreeRatio / numMDPTimeStepPerSecond)
maxWolfSpeed = int(14.5 * distanceToVisualDegreeRatio / numMDPTimeStepPerSecond)
wolfSubtleties = [500, 11, 3.3, 1.83, 0.92, 0.31, 0.001] # 0, 30, 60, .. 180
if idx == -1:
initWolfSubtlety = np.random.choice(wolfSubtleties)
else:
initWolfSubtlety = wolfSubtleties[idx]
prepareWolfVelocity = PrepareWolfVelocity(minWolfSpeed, maxWolfSpeed, warmUpTimeSteps, initWolfSubtlety,
transCartesianToPolar, transPolarToCartesian)
minDistractorSpeed = int(8.7 * distanceToVisualDegreeRatio / numMDPTimeStepPerSecond)
maxDistractorSpeed = int(14.5 * distanceToVisualDegreeRatio / numMDPTimeStepPerSecond)
prepareDistractorVelocity = PrepareDistractorVelocity(minDistractorSpeed, maxDistractorSpeed, warmUpTimeSteps,
transCartesianToPolar, transPolarToCartesian)
sheepId = 0
wolfId = 1
distractorsIds = list(range(2, numOfAgent))
prepareAllAgentsVelocities = PrepareAllAgentsVelocities(sheepId, wolfId, distractorsIds, prepareSheepVelocity,
prepareWolfVelocity, prepareDistractorVelocity)
xBoundary = [0, 640]
yBoundary = [0, 480]
stayInBoundaryByReflectVelocity = StayInBoundaryByReflectVelocity(xBoundary, yBoundary)
killzoneRadius = 2.5 * distanceToVisualDegreeRatio
isTerminal = IsTerminal(sheepId, wolfId, killzoneRadius)
numFramePerSecond = 30 # visual display fps
numFramesToInterpolate = int(
numFramePerSecond / numMDPTimeStepPerSecond - 1) # interpolate each MDP timestep to multiple frames; check terminal for each frame
transitFunction = TransitWithInterpolation(initWolfSubtlety, numFramesToInterpolate, prepareAllAgentsVelocities,
stayInBoundaryByReflectVelocity, isTerminal)
aliveBonus = 0.01
deathPenalty = -1
rewardFunction = RewardFunctionTerminalPenalty(aliveBonus, deathPenalty, isTerminal)
forwardOneStep = ForwardOneStep(transitFunction, rewardFunction)
return transitFunction, rewardFunction, forwardOneStep
def initializeEnvironment(numOfAgent):
sheepId = 0
wolfId = 1
distractorsIds = list(range(2, numOfAgent))
distanceToVisualDegreeRatio = 20
minInitSheepWolfDistance = 9 * distanceToVisualDegreeRatio
minInitSheepDistractorDistance = 2.5 * distanceToVisualDegreeRatio # no distractor in killzone when init
isLegalInitPositions = IsLegalInitPositions(sheepId, wolfId, distractorsIds, minInitSheepWolfDistance,
minInitSheepDistractorDistance)
xBoundary = [0, 640]
yBoundary = [0, 480]
resetState = ResetState(xBoundary, yBoundary, numOfAgent, isLegalInitPositions, transPolarToCartesian)
killzoneRadius = 2.5 * distanceToVisualDegreeRatio
isTerminal = IsTerminal(sheepId, wolfId, killzoneRadius)
return resetState,isTerminal
def __call__(self, parameters):
numOfAgent = parameters['numOfAgent']
trajectories = []
for trajectoryId in range(self.numTrajectories):
forwardOneStep = self.composeFowardOneTimeStepWithRandomSubtlety(
numOfAgent)
sheepId = 0
wolfId = 1
distractorsIds = list(range(2, numOfAgent))
distanceToVisualDegreeRatio = 20
minInitSheepWolfDistance = 9 * distanceToVisualDegreeRatio
minInitSheepDistractorDistance = 2.5 * distanceToVisualDegreeRatio # no distractor in killzone when init
isLegalInitPositions = IsLegalInitPositions(
sheepId, wolfId, distractorsIds, minInitSheepWolfDistance,
minInitSheepDistractorDistance)
xBoundary = [0, 600]
yBoundary = [0, 600]
resetState = ResetState(xBoundary, yBoundary, numOfAgent,
isLegalInitPositions,
transPolarToCartesian)
killzoneRadius = 2.5 * distanceToVisualDegreeRatio
isTerminal = IsTerminal(sheepId, wolfId, killzoneRadius)
numMDPTimeStepPerSecond = 5
maxRunningSteps = 25 * numMDPTimeStepPerSecond
sampleTrajecoty = SampleTrajectory(maxRunningSteps, isTerminal,
resetState, forwardOneStep)
numActionDirections = 8
actionSpace = [
(np.cos(directionId * 2 * math.pi / numActionDirections),
np.sin(directionId * 2 * math.pi / numActionDirections))
for directionId in range(numActionDirections)
]
randomPolicy = RandomPolicy(actionSpace)
sampleAction = lambda state: sampleFromDistribution(
randomPolicy(state))
trajectory = sampleTrajecoty(sampleAction)
trajectories.append(trajectory)
return trajectories
def main():
manipulatedVariables = OrderedDict()
manipulatedVariables['numOfAgent'] = [2]
levelNames = list(manipulatedVariables.keys())
levelValues = list(manipulatedVariables.values())
modelIndex = pd.MultiIndex.from_product(levelValues, names=levelNames)
toSplitFrame = pd.DataFrame(index=modelIndex)
productedValues = it.product(
*[[(key, value) for value in values]
for key, values in manipulatedVariables.items()])
parametersAllCondtion = [
dict(list(specificValueParameter))
for specificValueParameter in productedValues
]
numTrajectories = 3
sampleTrajectoriesForConditions = SampleTrajectoriesForCoditions(
numTrajectories, composeFowardOneTimeStepWithRandomSubtlety)
trajectoriesMultipleConditions = [
sampleTrajectoriesForConditions(para) for para in parametersAllCondtion
]
visualConditionIndex = 0
trajectoriesToVisualize = trajectoriesMultipleConditions[
visualConditionIndex]
visualize = True
if visualize:
screenWidth = 640
screenHeight = 480
screen = pg.display.set_mode((screenWidth, screenHeight))
screenColor = THECOLORS['black']
xBoundary = [0, 640]
yBoundary = [0, 480]
lineColor = THECOLORS['white']
lineWidth = 4
drawBackground = DrawBackground(screen, screenColor, xBoundary,
yBoundary, lineColor, lineWidth)
numOfAgent = 2
numDistractors = numOfAgent - 2
circleColorSpace = [[0, 255, 0], [255, 0, 0]
] + [[255, 255, 255]] * numDistractors
circleSize = 10
positionIndex = [0, 1]
agentIdsToDraw = list(range(numOfAgent))
saveImage = False
dirPYFile = os.path.dirname(__file__)
imageSavePath = os.path.join(dirPYFile, '..', 'data', 'forDemo')
if not os.path.exists(imageSavePath):
os.makedirs(imageSavePath)
FPS = 30
drawState = DrawState(FPS, screen, circleColorSpace, circleSize,
agentIdsToDraw, positionIndex, saveImage,
imageSavePath, drawBackground)
# MDP Env
xBoundary = [0, 640]
yBoundary = [0, 480]
stayInBoundaryByReflectVelocity = StayInBoundaryByReflectVelocity(
xBoundary, yBoundary)
distanceToVisualDegreeRatio = 20
killzoneRadius = 2.5 * distanceToVisualDegreeRatio
sheepId = 0
wolfId = 1
isTerminal = IsTerminal(sheepId, wolfId, killzoneRadius)
numMDPTimeStepPerSecond = 5 # change direction every 200ms
numFramesToInterpolate = int(
FPS / numMDPTimeStepPerSecond - 1
) # interpolate each MDP timestep to multiple frames; check terminal for each frame
interpolateStateForVisualization = InterpolateStateForVisualization(
numFramesToInterpolate, stayInBoundaryByReflectVelocity,
isTerminal)
stateIndexInTimeStep = 0
actionIndexInTimeStep = 1
nextStateIndexInTimeStep = 2
visualizeTraj = VisualizeTraj(stateIndexInTimeStep,
actionIndexInTimeStep,
nextStateIndexInTimeStep, drawState,
interpolateStateForVisualization)
[visualizeTraj(trajectory) for trajectory in trajectoriesToVisualize]