fix: this simulation and fifo work perfectly

Author: Cbampeta

src/Simulateur/WebotsSimulationGymEnvironment.py

@@ -18,6 +18,8 @@ class WebotsSimulationGymEnvironment(gym.Env):
     """
 
     def __init__(self, simulation_rank: int, vehicle_rank: int):
+
+        log("vvvvvvvvvvvvvvvvvvvvvvvvvv Initialisation vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv")
         super().__init__()
         self.simulation_rank = simulation_rank
         self.vehicle_rank = vehicle_rank
@@ -39,16 +41,17 @@ def __init__(self, simulation_rank: int, vehicle_rank: int):
         os.mkfifo(f"/tmp/autotech/{simulation_rank}_{vehicle_rank}tosupervisor.pipe")
 
         #  --mode=fast --minimize --no-rendering --batch --stdout
-        os.system(f"""
-            webots {__file__.rsplit('/', 1)[0]}/worlds/piste{simulation_rank % n_map}.wbt --mode=fast --minimize --no-rendering --batch --stdout &
-            echo $! {simulation_rank}_{vehicle_rank} >>/tmp/autotech/simulationranks
-        """)
+        if vehicle_rank == 0 :
+            os.system(f"""
+                webots {__file__.rsplit('/', 1)[0]}/worlds/piste{simulation_rank % n_map}.wbt --mode=fast --minimize --batch --stdout &
+                echo $! {simulation_rank}_{vehicle_rank} >>/tmp/autotech/simulationranks
+            """)
         log(f"SERVER{simulation_rank}_{vehicle_rank} : serverto{simulation_rank}_{vehicle_rank}.pipe")
         self.fifo_w = open(f"/tmp/autotech/serverto{simulation_rank}_{vehicle_rank}.pipe", "wb")
         log(f"SERVER{simulation_rank}_{vehicle_rank} : {simulation_rank}_{vehicle_rank}toserver.pipe")
         self.fifo_r = open(f"/tmp/autotech/{simulation_rank}_{vehicle_rank}toserver.pipe", "rb")
 
-        log("-------------------------------------------------------------------")
+        log("----------------------------- Inititalisation ---------------------------------")
 
     def reset(self, seed=0):
         # basically useless function
@@ -61,6 +64,8 @@ def reset(self, seed=0):
         return obs, info
 
     def step(self, action):
+
+        log("vvvvvvvvvvvvvvvvvvvvvvvvvv STEP vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv")
         log(f"SERVER{self.simulation_rank}_{self.vehicle_rank} : sending {action=}")
         self.fifo_w.write(action.tobytes())
         self.fifo_w.flush()
@@ -88,6 +93,7 @@ def step(self, action):
         ], axis=1)
 
         log(f"SERVER{self.simulation_rank}_{self.vehicle_rank} : step over")
+        log("----------------------------- STEP ---------------------------------")
 
         return obs, reward, done, truncated, info
 

src/Simulateur/config.py

@@ -3,7 +3,7 @@
 
 n_map = 2
 n_simulations = 2
-n_vehicles = 1
+n_vehicles = 2
 n_stupid_vehicles = 0
 n_actions_steering = 16
 n_actions_speed = 16

src/Simulateur/controllers/controllerVehicleDriver/controllerVehicleDriver.py

@@ -109,7 +109,11 @@ def step(self):
         # sends observation to the supervisor
 
         # First to be executed
-        self.fifo_w.write(self.observe().tobytes())
+        log(f"CLIENT{self.simulation_rank}/{self.i} : trying to write obs")
+        obs = self.observe()
+        log(f"CLIENT{self.simulation_rank}/{self.i} : driver sending {obs=}")
+        self.fifo_w.write(obs.tobytes())
+        self.fifo_w.flush() 
 
         log(f"CLIENT{self.simulation_rank}/{self.i} : trying to read from fifo")    
         action = np.frombuffer(self.fifo_r.read(np.dtype(np.int64).itemsize * 2), dtype=np.int64)

src/Simulateur/controllers/controllerWorldSupervisor/controllerWorldSupervisor.py

@@ -129,17 +129,22 @@ def __init__(self, vehicle_rank: int):
     # returns the lidar data of all vehicles
     def observe(self):
         # gets from Vehicle
-
-        return np.frombuffer(self.fifo_r.read(np.dtype(np.float32).itemsize * (n_sensors + lidar_horizontal_resolution + camera_horizontal_resolution)), dtype=np.float32)
+        log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : we observe")
+        obs = np.frombuffer(self.fifo_r.read(np.dtype(np.float32).itemsize * (n_sensors + lidar_horizontal_resolution + camera_horizontal_resolution)), dtype=np.float32)
+        log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : observing {obs=}")
+        return obs
+    
 
     # reset the gym environment reset
     def reset(self, seed=None):
+        log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : trying to reset vehicle")
         # this has to be done otherwise thec cars will shiver for a while sometimes when respawning and idk why
         if supervisor.getTime() - self.last_reset >= 1e-1:
+            log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : getting info")
             self.last_reset = supervisor.getTime()
 
             vehicle = supervisor.getFromDef(f"TT02_{self.vehicle_rank}")
-
+            log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : resetting all")
             self.checkpoint_manager.reset(seed)
             trans = self.checkpoint_manager.getTranslation()
             rot = self.checkpoint_manager.getRotation()
@@ -149,8 +154,9 @@ def reset(self, seed=None):
             self.checkpoint_manager.update()
 
             vehicle.resetPhysics()
+            log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : ok it's cooooooooool")
 
-        obs = self.observe()
+        obs = np.zeros(n_sensors + lidar_horizontal_resolution + camera_horizontal_resolution, dtype=np.float32)
         info = {}
         log(f"CLIENT{simulation_rank}/{self.vehicle_rank} : reset over")
         return obs, info
@@ -162,15 +168,17 @@ def step(self):
 
         # we should add a beacon sensor pointing upwards to detect the beacon
         obs = self.observe()
+        log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : observed {obs=}")
         sensor_data = obs[:n_sensors]
-
+        log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : sensor data {sensor_data=}")
         reward = 0
         done = np.False_
         truncated = np.False_
 
         x, y, z = self.translation_field.getSFVec3f()
         b_past_checkpoint = self.checkpoint_manager.update(x, y)
         b_collided, = sensor_data # unpack sensor data
+        log(f"SUPERVISOR{simulation_rank}/{self.vehicle_rank} : be_collided {b_collided=}")
 
         if b_collided or (z < -10):
             #print(f"CLIENT{simulation_rank}/{self.vehicle_rank} : {b_collided=}, {z=}")
@@ -216,7 +224,7 @@ def main():
         #Prédiction pour séléctionner une action à partir de l"observation
         for e in envs:
             obs, reward, done, truncated, info = e.step()
-            if done:
+            if  done:
                 obs, info = e.reset()
 
             log(f"SUPERVISOR{simulation_rank}/{e.vehicle_rank} : sending {obs=}")
@@ -229,6 +237,8 @@ def main():
             e.fifo_w.write(truncated.tobytes())
             e.fifo_w.flush()
 
+            
+
 
         for i in range(n_stupid_vehicles):
             tr_field = supervisor.getFromDef(f"TT02_{n_vehicles + i}").getField("translation")