From 388822a00bd0cf19707bafa91da120a26f7a0180 Mon Sep 17 00:00:00 2001
From: "Egli Adrian (IT-SCI-API-PFI)" <adrian.egli@sbb.ch>
Date: Tue, 22 Dec 2020 10:57:08 +0100
Subject: [PATCH] Policy updated
---
.../deadlockavoidance_with_decision_agent.py | 4 +++-
reinforcement_learning/multi_agent_training.py | 10 +++++++---
utils/dead_lock_avoidance_agent.py | 6 +-----
3 files changed, 11 insertions(+), 9 deletions(-)
diff --git a/reinforcement_learning/deadlockavoidance_with_decision_agent.py b/reinforcement_learning/deadlockavoidance_with_decision_agent.py
index a1726f8..e9a6f8e 100644
--- a/reinforcement_learning/deadlockavoidance_with_decision_agent.py
+++ b/reinforcement_learning/deadlockavoidance_with_decision_agent.py
@@ -10,6 +10,8 @@ from utils.dead_lock_avoidance_agent import DeadLockAvoidanceAgent
class DeadLockAvoidanceWithDecisionAgent(HybridPolicy):
def __init__(self, env: RailEnv, state_size, action_size, learning_agent):
+ print(">> DeadLockAvoidanceWithDecisionAgent")
+ super(DeadLockAvoidanceWithDecisionAgent, self).__init__()
self.env = env
self.state_size = state_size
self.action_size = action_size
@@ -33,7 +35,7 @@ class DeadLockAvoidanceWithDecisionAgent(HybridPolicy):
if agent.status < RailAgentStatus.DONE:
agents_on_switch, agents_near_to_switch, _, _ = \
self.agent_can_choose_helper.check_agent_decision(position, direction)
- if agents_on_switch:
+ if agents_on_switch or agents_near_to_switch:
return self.learning_agent.act(handle, state, eps)
else:
act = self.dead_lock_avoidance_agent.act(handle, state, -1.0)
diff --git a/reinforcement_learning/multi_agent_training.py b/reinforcement_learning/multi_agent_training.py
index 2e22854..6734f9c 100755
--- a/reinforcement_learning/multi_agent_training.py
+++ b/reinforcement_learning/multi_agent_training.py
@@ -177,12 +177,16 @@ def train_agent(train_params, train_env_params, eval_env_params, obs_params):
policy = PPOPolicy(state_size, get_action_size(), use_replay_buffer=False, in_parameters=train_params)
if False:
policy = DeadLockAvoidanceAgent(train_env, get_action_size())
- if False:
- inter_policy = PPOPolicy(state_size, get_action_size(), use_replay_buffer=False, in_parameters=train_params)
- policy = DeadLockAvoidanceWithDecisionAgent(train_env, state_size, get_action_size(), inter_policy)
if True:
+ inter_policy = DDDQNPolicy(state_size, get_action_size(), train_params)
+ policy = DeadLockAvoidanceWithDecisionAgent(train_env, state_size, get_action_size(), inter_policy)
+ if False:
policy = MultiDecisionAgent(state_size, get_action_size(), train_params)
+ # make sure that at least one policy is set
+ if policy is None:
+ policy = DDDQNPolicy(state_size, get_action_size(), train_params)
+
# Load existing policy
if train_params.load_policy is not "":
policy.load(train_params.load_policy)
diff --git a/utils/dead_lock_avoidance_agent.py b/utils/dead_lock_avoidance_agent.py
index ed3a3f7..ac7fd0c 100644
--- a/utils/dead_lock_avoidance_agent.py
+++ b/utils/dead_lock_avoidance_agent.py
@@ -6,7 +6,7 @@ from flatland.core.env_observation_builder import DummyObservationBuilder
from flatland.envs.agent_utils import RailAgentStatus
from flatland.envs.rail_env import RailEnv, RailEnvActions, fast_count_nonzero
-from reinforcement_learning.policy import HeuristicPolicy
+from reinforcement_learning.policy import HeuristicPolicy, DummyMemory
from utils.agent_action_config import map_rail_env_action
from utils.shortest_distance_walker import ShortestDistanceWalker
@@ -66,10 +66,6 @@ class DeadlockAvoidanceShortestDistanceWalker(ShortestDistanceWalker):
self.shortest_distance_agent_map[(handle, position[0], position[1])] = 1
self.full_shortest_distance_agent_map[(handle, position[0], position[1])] = 1
-
-
-
-
class DeadLockAvoidanceAgent(HeuristicPolicy):
def __init__(self, env: RailEnv, action_size, show_debug_plot=False):
self.env = env
--
GitLab