Changed in_range obervation space variable to visited

lsy_drone_racing/envs/drone_racing_deploy_env.py

@@ -92,9 +92,9 @@ def __init__(self, config: dict | Munch):
                 "target_gate": spaces.Discrete(n_gates, start=-1),
                 "gates_pos": spaces.Box(low=-np.inf, high=np.inf, shape=(n_gates, 3)),
                 "gates_rpy": spaces.Box(low=-np.pi, high=np.pi, shape=(n_gates, 3)),
-                "gates_in_range": spaces.Box(low=0, high=1, shape=(n_gates,), dtype=np.bool_),
+                "gates_visited": spaces.Box(low=0, high=1, shape=(n_gates,), dtype=np.bool_),
                 "obstacles_pos": spaces.Box(low=-np.inf, high=np.inf, shape=(n_obstacles, 3)),
-                "obstacles_in_range": spaces.Box(
+                "obstacles_visited": spaces.Box(
                     low=0, high=1, shape=(n_obstacles,), dtype=np.bool_
                 ),
             }
@@ -259,14 +259,14 @@ def obs(self) -> dict:
             self.gates_visited = np.logical_or(self.gates_visited, in_range)
             gates_pos[self.gates_visited] = real_gates_pos[self.gates_visited]
             gates_rpy[self.gates_visited] = real_gates_rpy[self.gates_visited]
-            obs["gates_in_range"] = in_range
+            obs["gates_visited"] = in_range
 
             in_range = (
                 np.linalg.norm(real_obstacles_pos[:, :2] - drone_pos[:2], axis=1) < sensor_range
             )
             self.obstacles_visited = np.logical_or(self.obstacles_visited, in_range)
             obstacles_pos[self.obstacles_visited] = real_obstacles_pos[self.obstacles_visited]
-            obs["obstacles_in_range"] = in_range
+            obs["obstacles_visited"] = in_range
 
         obs["gates_pos"] = gates_pos.astype(np.float32)
         obs["gates_rpy"] = gates_rpy.astype(np.float32)

lsy_drone_racing/envs/drone_racing_env.py

@@ -71,9 +71,9 @@ class DroneRacingEnv(gymnasium.Env):
     - "ang_vel": Drone angular velocity
     - "gates.pos": Positions of the gates
     - "gates.rpy": Orientations of the gates
-    - "gates.in_range": Flags indicating if the drone is in the sensor range of the gates
+    - "gates.visited": Flags indicating if the drone already was/ is in the sensor range of the gates and the true position is known
     - "obstacles.pos": Positions of the obstacles
-    - "obstacles.in_range": Flags indicating if the drone is in the sensor range of the obstacles
+    - "obstacles.visited": Flags indicating if the drone already was/ is in the sensor range of the obstacles and the true position is known
     - "target_gate": The current target gate index
 
     The action space consists of a desired full-state command
@@ -113,9 +113,9 @@ def __init__(self, config: dict):
                 "target_gate": spaces.Discrete(n_gates, start=-1),
                 "gates_pos": spaces.Box(low=-np.inf, high=np.inf, shape=(n_gates, 3)),
                 "gates_rpy": spaces.Box(low=-np.pi, high=np.pi, shape=(n_gates, 3)),
-                "gates_in_range": spaces.Box(low=0, high=1, shape=(n_gates,), dtype=np.bool_),
+                "gates_visited": spaces.Box(low=0, high=1, shape=(n_gates,), dtype=np.bool_),
                 "obstacles_pos": spaces.Box(low=-np.inf, high=np.inf, shape=(n_obstacles, 3)),
-                "obstacles_in_range": spaces.Box(
+                "obstacles_visited": spaces.Box(
                     low=0, high=1, shape=(n_obstacles,), dtype=np.bool_
                 ),
             }
@@ -125,6 +125,9 @@ def __init__(self, config: dict):
         self._steps = 0
         self._last_drone_pos = np.zeros(3)
 
+        self.gates_visited = np.array([False] * len(config.env.track.gates))
+        self.obstacles_visited = np.array([False] * len(config.env.track.obstacles))
+
     def reset(
         self, *, seed: int | None = None, options: dict | None = None
     ) -> tuple[dict[str, NDArray[np.floating]], dict]:
@@ -224,20 +227,30 @@ def obs(self) -> dict[str, NDArray[np.floating]]:
         # Add the gate and obstacle poses to the info. If gates or obstacles are in sensor range,
         # use the actual pose, otherwise use the nominal pose.
         in_range = self.sim.in_range(gates, self.sim.drone, self.config.env.sensor_range)
+        self.gates_visited = np.logical_or(self.gates_visited, in_range)
+
         gates_pos = np.stack([g["nominal.pos"] for g in gates.values()])
-        gates_pos[in_range] = np.stack([g["pos"] for g in gates.values()])[in_range]
+        gates_pos[self.gates_visited] = np.stack([g["pos"] for g in gates.values()])[
+            self.gates_visited
+        ]
         gates_rpy = np.stack([g["nominal.rpy"] for g in gates.values()])
-        gates_rpy[in_range] = np.stack([g["rpy"] for g in gates.values()])[in_range]
+        gates_rpy[self.gates_visited] = np.stack([g["rpy"] for g in gates.values()])[
+            self.gates_visited
+        ]
         obs["gates_pos"] = gates_pos.astype(np.float32)
         obs["gates_rpy"] = gates_rpy.astype(np.float32)
-        obs["gates_in_range"] = in_range
+        obs["gates_visited"] = self.gates_visited
 
         obstacles = self.sim.obstacles
         in_range = self.sim.in_range(obstacles, self.sim.drone, self.config.env.sensor_range)
+        self.obstacles_visited = np.logical_or(self.obstacles_visited, in_range)
+
         obstacles_pos = np.stack([o["nominal.pos"] for o in obstacles.values()])
-        obstacles_pos[in_range] = np.stack([o["pos"] for o in obstacles.values()])[in_range]
+        obstacles_pos[self.obstacles_visited] = np.stack([o["pos"] for o in obstacles.values()])[
+            self.obstacles_visited
+        ]
         obs["obstacles_pos"] = obstacles_pos.astype(np.float32)
-        obs["obstacles_in_range"] = in_range
+        obs["obstacles_visited"] = self.obstacles_visited
 
         if "observation" in self.sim.disturbances:
             obs = self.sim.disturbances["observation"].apply(obs)