simple_example.py is unstable

[OrionZou]

I run smarts.env:hiway-v0. it will stop unexpectedly.

my error (I clone SMARTS in Nov. 25 ,2020)

mycode：

import os

os.environ['SUMO_HOME'] = "/home/user/repos/sumo"

import gym
from smarts.core.agent_interface import AgentInterface, AgentType
from smarts.core.agent import AgentSpec, Agent
from elegantRL.env_util.continuous_space import agent_interface, observation_adapter, reward_adapter, action_adapter, \
    info_adapter, ACTION_SPACE
import numpy as np


class SimpleAgent(Agent):
    def __init__(self):
        self.action_space = gym.spaces.Box(
            low=np.array([0, 0, -1.0]), high=np.array([1.0, 1.0, 1.0]), dtype=np.float32
        )

    def act(self, obs):
        # return "keep_lane"
        return self.action_space.sample()


agent_spec = AgentSpec(
    interface=AgentInterface.from_type(AgentType.Full, max_episode_steps=None),
    agent_builder=SimpleAgent,
)

agent_specs = {
    "Agent-001": agent_spec,
    # "Agent-008": agent_spec,
}

env = gym.make(
    "smarts.env:hiway-v0",
    scenarios=["scenarios/loop"],
    agent_specs=agent_specs,
)

agents = {
    agent_id: agent_spec.build_agent()
    for agent_id, agent_spec in agent_specs.items()
}
observations = env.reset()

for i in range(100000000):
    print(i)
    agent_actions = {
        agent_id: agents[agent_id].act(agent_obs)
        for agent_id, agent_obs in observations.items()
    }
    for agent_id, agent_obs in observations.items():
        a = agents[agent_id].act(agent_obs)
    observations, r, d, i = env.step(agent_actions)
    if d:
        observations = env.reset()

[Gamenot]

@GyChou Hello, sorry for the late reply. We are currently dealing with a non-trivial edge case issue regarding SUMO.

#472

What this means in this case is that you will be unable to make use of bubbles for now. Since "scenarios/loop" contains a bubble it has the possibility to crash.

It will mean removing the bubble in scenarios/loop/scenario.py and then using scl scenario build scenarios/loop --clean to rebuild the scenario.

There is the potential that this is from a different source so I am going to run your example overnight to see if I can catch it.

[Gamenot]

Could you also update your post with your SUMO version and version of SMARTS?

You can get the sumo version with sumo --version (or /home/user/repos/sumo/bin/sumo in your case) and smarts version from your virtual environment with pip freeze | grep smarts if you used a package or by checking the version in setup.py if you are using the repository.

Thanks~

[Gamenot]

This crash is interesting, I have not seen it before. It appears like it could be from the number of times resetting SUMO.

@liamchzh has the behaviour with resetting SUMO changed? I seem to remember terminating and restarting SUMO before.

[liamchzh]

Yes, if we are reusing the same map when resetting, then reload SUMO instead of restart

SMARTS/smarts/core/sumo_traffic_simulation.py

Lines 230 to 234 in 151c748

	# restart sumo process only when map file changes
	if self._scenario and self._scenario.net_file_hash == scenario.net_file_hash:
	restart_sumo = False
	else:
	restart_sumo = True

[Gamenot]

I have confirmed, SUMO can freeze when reloading too many times.

pybullet build time: Dec 23 2020 01:47:29
32427
pybullet build time: Dec 23 2020 01:47:29
32428
pybullet build time: Dec 23 2020 01:47:29
32429
Traceback (most recent call last):
  File "examples/simple_example_test.py", line 55, in <module>
    observations = env.reset()
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/env/hiway_env.py", line 188, in reset
    env_observations = self._smarts.reset(scenario)
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/smarts.py", line 314, in reset
    self.setup(scenario)
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/smarts.py", line 361, in setup
    provider_state = self._setup_providers(self._scenario)
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/smarts.py", line 656, in _setup_providers
    provider_state.merge(provider.setup(scenario))
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/sumo_traffic_simulation.py", line 247, in setup
    self._traci_conn.simulation.subscribe(
  File "/usr/share/sumo/tools/traci/_simulation.py", line 468, in subscribe
    Domain.subscribe(self, "", varIDs, begin, end)
  File "/usr/share/sumo/tools/traci/domain.py", line 209, in subscribe
    self._connection._subscribe(self._subscribeID, begin, end, objectID, varIDs, parameters)
  File "/usr/share/sumo/tools/traci/connection.py", line 242, in _subscribe
    result = self._sendCmd(cmdID, (begin, end), objID, format, *args)
  File "/usr/share/sumo/tools/traci/connection.py", line 180, in _sendCmd
    return self._sendExact()
  File "/usr/share/sumo/tools/traci/connection.py", line 90, in _sendExact
    raise FatalTraCIError("connection closed by SUMO")
traci.exceptions.FatalTraCIError: connection closed by SUMO

My device also ended with a surprising error:

[Gamenot]

    if d:
        observations = env.reset()

One thing to note here is that this snippet should be:

    if d["__all__"]:
        observations = env.reset()

Or the example will end early.

[liamchzh]

I have confirmed, SUMO can freeze when reloading too many times.

pybullet build time: Dec 23 2020 01:47:29
32427
pybullet build time: Dec 23 2020 01:47:29
32428
pybullet build time: Dec 23 2020 01:47:29
32429
Traceback (most recent call last):
  File "examples/simple_example_test.py", line 55, in <module>
    observations = env.reset()
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/env/hiway_env.py", line 188, in reset
    env_observations = self._smarts.reset(scenario)
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/smarts.py", line 314, in reset
    self.setup(scenario)
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/smarts.py", line 361, in setup
    provider_state = self._setup_providers(self._scenario)
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/smarts.py", line 656, in _setup_providers
    provider_state.merge(provider.setup(scenario))
  File "/home/dev/Desktop/repos/NewSMARTS/SMARTS/smarts/core/sumo_traffic_simulation.py", line 247, in setup
    self._traci_conn.simulation.subscribe(
  File "/usr/share/sumo/tools/traci/_simulation.py", line 468, in subscribe
    Domain.subscribe(self, "", varIDs, begin, end)
  File "/usr/share/sumo/tools/traci/domain.py", line 209, in subscribe
    self._connection._subscribe(self._subscribeID, begin, end, objectID, varIDs, parameters)
  File "/usr/share/sumo/tools/traci/connection.py", line 242, in _subscribe
    result = self._sendCmd(cmdID, (begin, end), objID, format, *args)
  File "/usr/share/sumo/tools/traci/connection.py", line 180, in _sendCmd
    return self._sendExact()
  File "/usr/share/sumo/tools/traci/connection.py", line 90, in _sendExact
    raise FatalTraCIError("connection closed by SUMO")
traci.exceptions.FatalTraCIError: connection closed by SUMO

Are you able to reproduce this if we force SUMO to restart everytime?

[Gamenot]

Closed since we found the cause in #480.