Tests envs (#235)

* testing lane change env is working

* empty tests for environments

* fixed TestEnv and added tests

* fixed 0 initial vehicles bug

* added tests for keyerrors

* pep8

* bug fixes

* more tests to envs

* pep8

* converted tests into utility methods

examples/sumo/highway.py

@@ -5,7 +5,8 @@
 from flow.core.params import SumoParams, EnvParams, \
     NetParams, InitialConfig, InFlows
 from flow.core.vehicles import Vehicles
-from flow.envs.loop.loop_accel import AccelEnv, ADDITIONAL_ENV_PARAMS
+from flow.envs.loop.lane_changing import LaneChangeAccelEnv, \
+    ADDITIONAL_ENV_PARAMS
 from flow.scenarios.highway import HighwayScenario, ADDITIONAL_NET_PARAMS
 
 
@@ -67,7 +68,7 @@ def highway_example(render=None):
         net_params=net_params,
         initial_config=initial_config)
 
-    env = AccelEnv(env_params, sumo_params, scenario)
+    env = LaneChangeAccelEnv(env_params, sumo_params, scenario)
 
     return SumoExperiment(env, scenario)
 

flow/envs/loop/lane_changing.py

@@ -240,14 +240,13 @@ def get_state(self):
                     lane_headways[j] /= max_length
                     vel_in_front[j] = self.vehicles.get_speed(lane_leader) \
                         / max_speed
+                    self.visible.extend([lane_leader])
             for j, lane_follower in enumerate(lane_followers):
                 if lane_follower != '':
                     lane_headways[j] /= max_length
                     vel_behind[j] = self.vehicles.get_speed(lane_follower) \
                         / max_speed
-
-            self.visible.extend(lane_leaders)
-            self.visible.extend(lane_followers)
+                    self.visible.extend([lane_follower])
 
             # add the headways, tailways, and speed for all lane leaders
             # and followers

flow/envs/loop/loop_accel.py

@@ -4,7 +4,6 @@
 from flow.core import rewards
 
 from gym.spaces.box import Box
-from gym.spaces.tuple_space import Tuple
 
 import numpy as np
 
@@ -68,18 +67,11 @@ def action_space(self):
     @property
     def observation_space(self):
         """See class definition."""
-        self.obs_var_labels = ["Velocity", "Absolute_pos"]
-        speed = Box(
+        return Box(
             low=0,
             high=1,
-            shape=(self.vehicles.num_vehicles, ),
-            dtype=np.float32)
-        pos = Box(
-            low=0.,
-            high=1,
-            shape=(self.vehicles.num_vehicles, ),
+            shape=(2 * self.vehicles.num_vehicles, ),
             dtype=np.float32)
-        return Tuple((speed, pos))
 
     def _apply_rl_actions(self, rl_actions):
         """See class definition."""
@@ -98,13 +90,12 @@ def compute_reward(self, rl_actions, **kwargs):
 
     def get_state(self, **kwargs):
         """See class definition."""
-        # speed normalizer
-        max_speed = self.scenario.max_speed
+        speed = [self.vehicles.get_speed(veh_id) / self.scenario.max_speed
+                 for veh_id in self.sorted_ids]
+        pos = [self.get_x_by_id(veh_id) / self.scenario.length
+               for veh_id in self.sorted_ids]
 
-        return np.array([[
-            self.vehicles.get_speed(veh_id) / max_speed,
-            self.get_x_by_id(veh_id) / self.scenario.length
-        ] for veh_id in self.sorted_ids])
+        return np.array(speed + pos)
 
     def additional_command(self):
         """Define which vehicles are observed for visualization purposes."""

flow/envs/test.py

@@ -31,22 +31,21 @@ class and returns a real number.
 
     @property
     def action_space(self):
-        """See class definition."""
-        return Box(low=0, high=0, shape=(1,), dtype=np.float32)
+        return Box(low=0, high=0, shape=(0,), dtype=np.float32)
 
     @property
     def observation_space(self):
-        """See class definition."""
-        return Box(low=0, high=0, shape=(1,), dtype=np.float32)
+        return Box(low=0, high=0, shape=(0,), dtype=np.float32)
 
     def _apply_rl_actions(self, rl_actions):
-        """See class definition."""
         return
 
     def compute_reward(self, rl_actions, **kwargs):
-        """See class definition."""
-        return 0
+        if "reward_fn" in self.env_params.additional_params:
+            return self.env_params.additional_params["reward_fn"](self)
+        else:
+            return 0
 
     def get_state(self, **kwargs):
         """See class definition."""
-        return np.array([0])
+        return np.array([])

flow/scenarios/base_scenario.py

@@ -340,6 +340,11 @@ def gen_even_start_pos(self, initial_config, num_vehicles, **kwargs):
          available_edges, initial_config) = \
             self._get_start_pos_util(initial_config, num_vehicles, **kwargs)
 
+        # return an empty list of starting positions and lanes if there are no
+        # vehicles to be placed
+        if num_vehicles == 0:
+            return [], []
+
         increment = available_length / num_vehicles
 
         # if not all lanes are equal, then we must ensure that vehicles are in