checked out action_shaping from dev branch

examples/ecr/q_learning/common/action_shaping.py

@@ -9,7 +9,6 @@ from typing import Tuple
 from maro.simulator.scenarios.ecr.common import ActionScope
 from maro.utils import Logger, LogFormat
 
-
 class DiscreteActionShaping():
     def __init__(self, action_space: [float]):
         '''
@@ -41,7 +40,11 @@ class DiscreteActionShaping():
             return max(round(self._action_space[action_index] * port_empty), -vessel_remaining_space)
 
         if action_index > self._zero_action_index:
-            return round(self._action_space[action_index] * scope.discharge)
+            plan_action = self._action_space[action_index] * (scope.discharge + early_discharge) - early_discharge
+            if plan_action > 0:
+                return round(plan_action)
+            else:
+                return round(self._action_space[action_index] * scope.discharge) 
 
         return 0