From c9dc6e10d7d7880a13662dc1e9bcf3020f6f8004 Mon Sep 17 00:00:00 2001
From: MLErik <baerenjesus@gmail.com>
Date: Thu, 31 Oct 2019 14:20:36 -0400
Subject: [PATCH] refactored how we call malfuncitons. Now we draw random
numbers at each time step for each agent. will this be to expensive? If yes
this can easily be refactored
---
flatland/envs/rail_env.py | 76 +++++++++++++++++++++------------------
1 file changed, 42 insertions(+), 34 deletions(-)
diff --git a/flatland/envs/rail_env.py b/flatland/envs/rail_env.py
index 58faab24..82e49239 100644
--- a/flatland/envs/rail_env.py
+++ b/flatland/envs/rail_env.py
@@ -367,15 +367,17 @@ class RailEnv(Environment):
for i_agent in range(self.get_num_agents()):
self.set_agent_active(i_agent)
- # Induce malfunctions
- self._malfunction(self.mean_malfunction_rate)
+
for agent in self.agents:
+ # Induce malfunctions
+ self._break_agent(self.mean_malfunction_rate, agent)
+
if agent.malfunction_data["malfunction"] > 0:
agent.speed_data['transition_action_on_cellexit'] = RailEnvActions.DO_NOTHING
- # Fix agents that finished their malfunciton
- self._fix_agents()
+ # Fix agents that finished their malfunction
+ self._fix_agent(agent)
self.num_resets += 1
self._elapsed_steps = 0
@@ -399,42 +401,46 @@ class RailEnv(Environment):
observation_dict: Dict = self._get_observations()
return observation_dict, info_dict
- def _fix_agents(self):
+ def _fix_agent(self, agent):
"""
Updates agent malfunction variables and fixes broken agents
- """
- for agent in self.agents:
- # Ignore agents that OK
- if self._agent_is_ok(agent):
- continue
+ Parameters
+ ----------
+ agent
+ """
- # Reduce number of malfunction steps left
- if agent.malfunction_data['malfunction'] > 1:
- agent.malfunction_data['malfunction'] -= 1
- continue
+ # Ignore agents that are OK
+ if self._is_ok(agent):
+ return
- # Restart agents at the end of their malfunction
+ # Reduce number of malfunction steps left
+ if agent.malfunction_data['malfunction'] > 1:
agent.malfunction_data['malfunction'] -= 1
- if 'moving_before_malfunction' in agent.malfunction_data:
- agent.moving = agent.malfunction_data['moving_before_malfunction']
- continue
+ return
+ # Restart agents at the end of their malfunction
+ agent.malfunction_data['malfunction'] -= 1
+ if 'moving_before_malfunction' in agent.malfunction_data:
+ agent.moving = agent.malfunction_data['moving_before_malfunction']
+ return
- def _malfunction(self, rate):
+ def _break_agent(self, rate, agent):
"""
- Malfunction generator that breaks agents at a given rate. It does randomly chose agent to break during the run
+ Malfunction generator that breaks agents at a given rate.
- """
- if self.np_random.rand() < self._malfunction_prob(rate, len(self.active_agents)):
+ Parameters
+ ----------
+ agent
- breaking_agent = self._draw_malfunctioning_agent(0)
- if breaking_agent:
+ """
+ if agent.malfunction_data['malfunction'] < 1:
+ if self.np_random.rand() < self._malfunction_prob(rate):
num_broken_steps = self.np_random.randint(self.min_number_of_steps_broken,
self.max_number_of_steps_broken + 1) + 1
- breaking_agent.malfunction_data['malfunction'] = num_broken_steps
- breaking_agent.malfunction_data['moving_before_malfunction'] = breaking_agent.moving
- breaking_agent.malfunction_data['nr_malfunctions'] += 1
+ agent.malfunction_data['malfunction'] = num_broken_steps
+ agent.malfunction_data['moving_before_malfunction'] = agent.moving
+ agent.malfunction_data['nr_malfunctions'] += 1
return
@@ -477,13 +483,15 @@ class RailEnv(Environment):
}
have_all_agents_ended = True # boolean flag to check if all agents are done
- # Induce malfunctions
- self._malfunction(self.mean_malfunction_rate)
+
for i_agent, agent in enumerate(self.agents):
# Reset the step rewards
self.rewards_dict[i_agent] = 0
+ # Induce malfunction before we do a step, thus a broken agent can't move in this step
+ self._break_agent(self.mean_malfunction_rate, agent)
+
# Perform step on the agent
self._step_agent(i_agent, action_dict_.get(i_agent))
@@ -496,8 +504,8 @@ class RailEnv(Environment):
info_dict["speed"][i_agent] = agent.speed_data['speed']
info_dict["status"][i_agent] = agent.status
- # Fix agents that finished their malfunction
- self._fix_agents()
+ # Fix agents that finished their malfunction such that they can perfom an action in the next step
+ self._fix_agent(agent)
# Check for end of episode + set global reward to all rewards!
if have_all_agents_ended:
@@ -960,7 +968,7 @@ class RailEnv(Environment):
x = - np.log(1 - u) * rate
return x
- def _malfunction_prob(self, rate, n_agents):
+ def _malfunction_prob(self, rate):
"""
Probability that an agent break given the number of agents an the probability of a sinlge agent to break
:param rate:
@@ -969,7 +977,7 @@ class RailEnv(Environment):
if rate <= 0:
return 0.
else:
- return 1 - np.exp(- (1 / rate) * (n_agents))
+ return 1 - np.exp(- (1 / rate))
def _draw_malfunctioning_agent(self, tries):
"""
@@ -1000,7 +1008,7 @@ class RailEnv(Environment):
else:
return breaking_agent
- def _agent_is_ok(self, agent):
+ def _is_ok(self, agent):
"""
Check if an agent is ok, meaning it can move and is not malfuncitoinig
Parameters
--
GitLab