diff --git a/.idea/inspectionProfiles/profiles_settings.xml b/.idea/inspectionProfiles/profiles_settings.xml index 105ce2da2d6447d11dfe32bfb846c3d5b199fc99..20fc29e7a85e1af4cdab6fc32d2197bfd9cc1d27 100644 --- a/.idea/inspectionProfiles/profiles_settings.xml +++ b/.idea/inspectionProfiles/profiles_settings.xml @@ -1,6 +1,6 @@ -<component name="InspectionProjectProfileManager"> - <settings> - <option name="USE_PROJECT_PROFILE" value="false" /> - <version value="1.0" /> - </settings> +<component name="InspectionProjectProfileManager"> + <settings> + <option name="USE_PROJECT_PROFILE" value="false" /> + <version value="1.0" /> + </settings> </component> \ No newline at end of file diff --git a/checkpoints/ppo/model_checkpoint.meta b/checkpoints/ppo/model_checkpoint.meta index 56a27a0763598ba9748c4b337fcb59e95ccdf612..9346fabc64acda708cfa4b7153867cd7e9cc908c 100644 Binary files a/checkpoints/ppo/model_checkpoint.meta and b/checkpoints/ppo/model_checkpoint.meta differ diff --git a/checkpoints/ppo/model_checkpoint.optimizer b/checkpoints/ppo/model_checkpoint.optimizer index ff9dec932ff15047548262d1c0a0e0f52a79cbc6..cbaca012f49cee8e248536632f37ee8734a65ecd 100644 Binary files a/checkpoints/ppo/model_checkpoint.optimizer and b/checkpoints/ppo/model_checkpoint.optimizer differ diff --git a/checkpoints/ppo/model_checkpoint.policy b/checkpoints/ppo/model_checkpoint.policy index f1258b047ae66c183f0c9d25b0b1b0cef274d9c9..43b03075f900b82fa4eac79dc21c9ce0e158a128 100644 Binary files a/checkpoints/ppo/model_checkpoint.policy and b/checkpoints/ppo/model_checkpoint.policy differ diff --git a/run.py b/run.py index 4eab22942b0914d57606a7226d6bd250c81401dd..a2f071cbbdd264f23e7f8eda56bf78fdaba4614e 100644 --- a/run.py +++ b/run.py @@ -30,7 +30,7 @@ def my_controller(extra: Extra, observation, info): # the example here : # https://gitlab.aicrowd.com/flatland/flatland/blob/master/flatland/envs/observations.py#L14 ##################################################################### -my_observation_builder = Extra(max_depth=3) +my_observation_builder = Extra(max_depth=1) # Or if you want to use your own approach to build the observation from the env_step, # please feel free to pass a DummyObservationBuilder() object as mentioned below, diff --git a/src/extra.py b/src/extra.py index b70830f91714c3a8a3efdabe34bd1a427ea1f73e..7049bc05dd5454b2fec68bee61ae69c3a92d8644 100644 --- a/src/extra.py +++ b/src/extra.py @@ -353,7 +353,12 @@ class Extra(ObservationBuilder): def rl_agent_act(self, observation, info, eps=0.0): if len(self.random_agent_starter) != self.env.get_num_agents(): - self.random_agent_starter = np.random.random(self.env.get_num_agents()) * 1000.0 + f = self.env._max_episode_steps + if f is None: + f = 1000.0 + else: + f *= 0.8 + self.random_agent_starter = np.random.random(self.env.get_num_agents()) * f self.loadAgent() action_dict = {}