def _assemble_reward(self):
self._explore_rewards = [
ir.EvadeInTargetRangeRwd(self._compress_width, self._range_width),
sr.ViewEnemyReward(weight=30)
]
self._backward_rewards = [br.BackwardRwd(), br.BackwardFinalRwd()]
self._final_rewards = [ir.EvadeUnderAttackRwd(), ir.EvadeFinalRwd()]
def _assemble_reward(self):
self._rewards = [
ir.EvadeUnderAttackRwd(),
ir.EvadeInTargetRangeRwd(self._compress_width, self._range_width),
sr.ViewEnemyReward(weight=30),
ir.EvadeFinalRwd()
]
def _assemble_reward(self):
self._rewards = [
ir.EvadeUnderAttackRwd(),
ir.EvadeFinalRwd(),
ir.EvadeInTargetRangeRwd(),
sr.ViewEnemyReward(weight=20)
]
def _assemble_reward(self):
self._rewards = [
ir.EvadeUnderAttackRwd(),
ir.EvadeInTargetRangeRwd(self._compress_width, self._range_width),
fr.FullGameViewRwd(weight=5),
ir.EvadeFinalRwd()
]
def _assemble_reward(self):
self._forward_rewards = [
sr.HomeReward(negative=True),
sr.EnemyBaseReward(negative=True),
sr.MinDistReward(negative=True)
]
self._backward_rewards = [
sr.HomeReward(negative=True, back=True),
sr.EnemyBaseReward(negative=True, back=True),
ir.EvadeUnderAttackRwd()
]
self._explore_rewards = [
ir.EvadeUnderAttackRwd(),
ir.EvadeInTargetRangeRwd(self._compress_width, self._range_width),
sr.ViewEnemyReward(weight=20)
#ir.EvadeTargetScanRwd(self._compress_width, self._range_width),
]
self._final_rewards = [ir.EvadeFinalRwd()]