From 9c66a35f82435836aebfb159763ce38f0954326c Mon Sep 17 00:00:00 2001
From: Erik Nygren <erik.nygren@sbb.ch>
Date: Tue, 23 Apr 2019 10:29:57 +0200
Subject: [PATCH] found bug --> Infinite loops that you can enter but never
leave...
---
examples/training_navigation.py | 3 ++-
flatland/core/env_observation_builder.py | 1 +
flatland/envs/rail_env.py | 3 ---
3 files changed, 3 insertions(+), 4 deletions(-)
diff --git a/examples/training_navigation.py b/examples/training_navigation.py
index 975d33fb..b78851c8 100644
--- a/examples/training_navigation.py
+++ b/examples/training_navigation.py
@@ -63,8 +63,9 @@ for trials in range(1, n_trials + 1):
action_dict.update({a: action})
# Environment step
+ print(trials,step)
next_obs, all_rewards, done, _ = env.step(action_dict)
-
+ print("stepped")
# Update replay buffer and train agent
for a in range(env.number_of_agents):
diff --git a/flatland/core/env_observation_builder.py b/flatland/core/env_observation_builder.py
index d7bee930..ce0a28ae 100644
--- a/flatland/core/env_observation_builder.py
+++ b/flatland/core/env_observation_builder.py
@@ -272,6 +272,7 @@ class TreeObsForRailEnv(ObservationBuilder):
other_agent_encountered = False
other_target_encountered = False
while exploring:
+
# #############################
# #############################
# Modify here to compute any useful data required to build the end node's features. This code is called
diff --git a/flatland/envs/rail_env.py b/flatland/envs/rail_env.py
index 36040d5c..37a97ff8 100644
--- a/flatland/envs/rail_env.py
+++ b/flatland/envs/rail_env.py
@@ -612,9 +612,7 @@ class RailEnv(Environment):
direction = reverse_direction
movement = reverse_direction
is_deadend = True
-
new_position = self._new_position(pos, movement)
-
# Is it a legal move? 1) transition allows the movement in the
# cell, 2) the new cell is not empty (case 0), 3) the cell is
# free, i.e., no agent is currently in that cell
@@ -668,7 +666,6 @@ class RailEnv(Environment):
# Reset the step actions (in case some agent doesn't 'register_action'
# on the next step)
self.actions = [0]*self.number_of_agents
-
return self._get_observations(), self.rewards_dict, self.dones, {}
def _new_position(self, position, movement):
--
GitLab