def get_reward(self, new_stats, old_stats):
#longer path is rewarded and less number of regions is rewarded
rewards = {
"regions": get_range_reward(new_stats["regions"], old_stats["regions"], 1, 1),
"path-length": get_range_reward(new_stats["path-length"],old_stats["path-length"], np.inf, np.inf)
}
#calculate the total reward
return rewards["regions"] * self._rewards["regions"] +\
rewards["path-length"] * self._rewards["path-length"]
def get_reward(self, new_stats, old_stats):
#longer path is rewarded and less number of regions is rewarded
rewards = {
"player":
get_range_reward(new_stats["player"], old_stats["player"], 1, 1),
"key":
get_range_reward(new_stats["key"], old_stats["key"], 1, 1),
"door":
get_range_reward(new_stats["door"], old_stats["door"], 1, 10),
"enemies":
get_range_reward(new_stats["enemies"], old_stats["enemies"], 2,
self._max_enemies),
"regions":
get_range_reward(new_stats["regions"], old_stats["regions"], 1, 1),
"nearest-enemy":
get_range_reward(new_stats["nearest-enemy"],
old_stats["nearest-enemy"],
self._target_enemy_dist, np.inf),
"path-length":
get_range_reward(new_stats["path-length"],
old_stats["path-length"], np.inf, np.inf)
}
#calculate the total reward
return rewards["player"] * self._rewards["player"] +\
rewards["key"] * self._rewards["key"] +\
rewards["door"] * self._rewards["door"] +\
rewards["enemies"] * self._rewards["enemies"] +\
rewards["regions"] * self._rewards["regions"] +\
rewards["nearest-enemy"] * self._rewards["nearest-enemy"] +\
rewards["path-length"] * self._rewards["path-length"]
def get_reward(self, new_stats, old_stats):
#longer path is rewarded and less number of regions is rewarded
rewards = {
"player":
get_range_reward(new_stats["player"], old_stats["player"], 1, 1),
"crate":
get_range_reward(new_stats["crate"], old_stats["crate"], 1,
self._max_crates),
"target":
get_range_reward(new_stats["target"], old_stats["target"], 1,
self._max_crates),
"regions":
get_range_reward(new_stats["regions"], old_stats["regions"], 1, 1),
"ratio":
get_range_reward(abs(new_stats["crate"] - new_stats["target"]),
abs(old_stats["crate"] - old_stats["target"]),
-np.inf, -np.inf),
"dist-win":
get_range_reward(new_stats["dist-win"], old_stats["dist-win"],
-np.inf, -np.inf),
"sol-length":
get_range_reward(len(new_stats["solution"]),
len(old_stats["solution"]), np.inf, np.inf)
}
#calculate the total reward
return rewards["player"] * self._rewards["player"] +\
rewards["crate"] * self._rewards["crate"] +\
rewards["target"] * self._rewards["target"] +\
rewards["regions"] * self._rewards["regions"] +\
rewards["ratio"] * self._rewards["ratio"] +\
rewards["dist-win"] * self._rewards["dist-win"] +\
rewards["sol-length"] * self._rewards["sol-length"]
def get_reward(self, new_stats, old_stats):
#longer path is rewarded and less number of regions is rewarded
rewards = {
"player":
get_range_reward(new_stats["player"], old_stats["player"], 1, 1),
"exit":
get_range_reward(new_stats["exit"], old_stats["exit"], 1, 1),
"diamonds":
get_range_reward(new_stats["diamonds"], old_stats["diamonds"],
-np.inf, self._max_diamonds),
"dist-floor":
get_range_reward(new_stats["dist-floor"], old_stats["dist-floor"],
0, 0),
"key":
get_range_reward(new_stats["key"], old_stats["key"], 1, 1),
"spikes":
get_range_reward(new_stats["spikes"], old_stats["spikes"],
self._min_spikes, np.inf),
"regions":
get_range_reward(new_stats["regions"], old_stats["regions"], 1, 1),
"num-jumps":
get_range_reward(new_stats["num-jumps"], old_stats["num-jumps"],
np.inf, np.inf),
"dist-win":
get_range_reward(new_stats["dist-win"], old_stats["dist-win"],
-np.inf, -np.inf),
"sol-length":
get_range_reward(new_stats["sol-length"], old_stats["sol-length"],
np.inf, np.inf)
}
#calculate the total reward
return rewards["player"] * self._rewards["player"] +\
rewards["dist-floor"] * self._rewards["dist-floor"] +\
rewards["exit"] * self._rewards["exit"] +\
rewards["spikes"] * self._rewards["spikes"] +\
rewards["diamonds"] * self._rewards["diamonds"] +\
rewards["key"] * self._rewards["key"] +\
rewards["regions"] * self._rewards["regions"] +\
rewards["num-jumps"] * self._rewards["num-jumps"] +\
rewards["dist-win"] * self._rewards["dist-win"] +\
rewards["sol-length"] * self._rewards["sol-length"]
def get_reward(self, new_stats, old_stats):
# longer path is rewarded and less number of regions is rewarded
rewards = {
"base_count":
get_range_reward(new_stats["base_count"], old_stats["base_count"],
self._target_base, self._target_base),
"base_distance":
get_range_reward(new_stats["base_distance"],
old_stats["base_distance"], self._width / 2,
self._width * 2),
# "base_space": get_range_reward(new_stats["base_space"], old_stats["base_space"], 0, 10),
# "asymmetry": get_range_reward(new_stats["asymmetry"], old_stats["asymmetry"], 0, 5),
"resource_count":
get_range_reward(new_stats["resource_count"],
old_stats["resource_count"], self._min_resource,
self._max_resource),
"resource_distance":
get_range_reward(new_stats["resource_distance"],
old_stats["resource_distance"], 0, 1),
# "resource_clustering": get_range_reward(new_stats["resource_clustering"], old_stats["resource_clustering"], 0, 5),
# "path_overlapping": get_range_reward(new_stats["path_overlapping"], old_stats["path_overlapping"], 0, 10),
"chock_point":
get_range_reward(new_stats["chock_point"],
old_stats["chock_point"], 0,
self._max_chock_points),
"region":
get_range_reward(new_stats["region"], old_stats["region"], 1, 1)
}
# calculate the total reward
return rewards["base_count"] * self._rewards["base_count"] + \
rewards["base_distance"] * self._rewards["base_distance"] + \
rewards["resource_count"] * self._rewards["resource_count"] + \
rewards["region"] * self._rewards["region"] + \
rewards["resource_distance"] * self._rewards["resource_distance"] + \
rewards["chock_point"] * self._rewards["chock_point"]
def get_reward(self, new_stats, old_stats):
#longer path is rewarded and less number of regions is rewarded
rewards = {
"player":
get_range_reward(new_stats["player"], old_stats["player"], 1, 1),
"exit":
get_range_reward(new_stats["exit"], old_stats["exit"], 1, 1),
"potions":
get_range_reward(new_stats["potions"], old_stats["potions"],
-np.inf, self._max_potions),
"treasures":
get_range_reward(new_stats["treasures"], old_stats["treasures"],
-np.inf, self._max_treasures),
"enemies":
get_range_reward(new_stats["enemies"], old_stats["enemies"], 1,
self._max_enemies),
"regions":
get_range_reward(new_stats["regions"], old_stats["regions"], 1, 1),
"col-enemies":
get_range_reward(new_stats["col-enemies"],
old_stats["col-enemies"], np.inf, np.inf),
"dist-win":
get_range_reward(new_stats["dist-win"], old_stats["dist-win"],
-np.inf, -np.inf),
"sol-length":
get_range_reward(new_stats["sol-length"], old_stats["sol-length"],
np.inf, np.inf)
}
#calculate the total reward
return rewards["player"] * self._rewards["player"] +\
rewards["exit"] * self._rewards["exit"] +\
rewards["enemies"] * self._rewards["enemies"] +\
rewards["treasures"] * self._rewards["treasures"] +\
rewards["potions"] * self._rewards["potions"] +\
rewards["regions"] * self._rewards["regions"] +\
rewards["col-enemies"] * self._rewards["col-enemies"] +\
rewards["dist-win"] * self._rewards["dist-win"] +\
rewards["sol-length"] * self._rewards["sol-length"]
def get_reward(self, new_stats, old_stats):
# longer path is rewarded and less number of regions is rewarded
rewards = {
"base_count":
get_range_reward(new_stats["base_count"], old_stats["base_count"],
self._target_base, self._target_base),
"base_distance":
get_range_reward(new_stats["base_distance"],
old_stats["base_distance"], 0,
self._base_distance_diff),
"resource_count":
get_range_reward(new_stats["resource_count"],
old_stats["resource_count"], self._min_resource,
self._max_resource),
"resource_distance":
get_range_reward(new_stats["resource_distance"],
old_stats["resource_distance"], 0,
self._resource_distance_diff),
"resource_balance":
get_range_reward(new_stats["resource_balance"],
old_stats["resource_balance"], 0,
self._resource_balance_diff),
"obstacle":
get_range_reward(new_stats["obstacle"], old_stats["obstacle"], 0,
self._max_obstacles),
"region":
get_range_reward(new_stats["region"], old_stats["region"], 1, 1),
"area_control":
get_range_reward(new_stats["area_control"],
old_stats["area_control"], 0,
self._area_control_diff),
}
# calculate the total reward
return rewards["base_count"] * self._rewards["base_count"] + \
rewards["base_distance"] * self._rewards["base_distance"] + \
rewards["resource_count"] * self._rewards["resource_count"] + \
rewards["region"] * self._rewards["region"] + \
rewards["resource_distance"] * self._rewards["resource_distance"] + \
rewards["resource_balance"] * self._rewards["resource_balance"] + \
rewards["obstacle"] * self._rewards["obstacle"] + \
rewards["area_control"] * self._rewards["area_control"]
def get_reward(self, new_stats, old_stats):
#longer path is rewarded and less number of regions is rewarded
rewards = {
"dist-floor": get_range_reward(new_stats["dist-floor"], old_stats["dist-floor"], 0, 0),
"disjoint-tubes": get_range_reward(new_stats["disjoint-tubes"], old_stats["disjoint-tubes"], 0, 0),
"enemies": get_range_reward(new_stats["enemies"], old_stats["enemies"], self._min_enemies, self._max_enemies),
"empty": get_range_reward(new_stats["empty"], old_stats["empty"], self._min_empty, np.inf),
"noise": get_range_reward(new_stats["noise"], old_stats["noise"], 0, 0),
"jumps": get_range_reward(new_stats["jumps"], old_stats["jumps"], self._min_jumps, np.inf),
"jumps-dist": get_range_reward(new_stats["jumps-dist"], old_stats["jumps-dist"], 0, 0),
"dist-win": get_range_reward(new_stats["dist-win"], old_stats["dist-win"], 0, 0)
}
#calculate the total reward
return rewards["dist-floor"] * self._rewards["dist-floor"] +\
rewards["disjoint-tubes"] * self._rewards["disjoint-tubes"] +\
rewards["enemies"] * self._rewards["enemies"] +\
rewards["empty"] * self._rewards["empty"] +\
rewards["noise"] * self._rewards["noise"] +\
rewards["jumps"] * self._rewards["jumps"] +\
rewards["jumps-dist"] * self._rewards["jumps-dist"] +\
rewards["dist-win"] * self._rewards["dist-win"]