feat(*): add staged curriculum training with an automated stage-by-stage orchestrator

Game/scripts/ship_ai_controller.gd

@@ -71,6 +71,16 @@ extends AIController3D
 # chance appears, instead of a free way to keep collecting dense reward.
 @export var time_penalty := 0.001
 
+# Locomotion curriculum: when false, the corresponding action axes are
+# discarded in set_action before reaching the ship, so the ship stays
+# grounded and only yaws — basic scoring/defending doesn't need 3D flight.
+# This masks the *effect* of thrust.y/rotation.x/rotation.z, not the action
+# space's shape: the policy still outputs values for these axes (still
+# contributing to PPO's entropy/log-prob), they're just discarded here, so
+# checkpoints stay resumable once a later curriculum stage re-enables them.
+@export var allow_vertical := true
+@export var allow_pitch_roll := true
+
 # Contact normals with y above this are floor contact (exempt from the wall
 # penalty); below it they read as wall (sideways) or ceiling (downward).
 const FLOOR_NORMAL_MIN_Y := 0.7
@@ -128,8 +138,11 @@ func get_action_space() -> Dictionary:
 func set_action(action) -> void:
 	var thrust: Array = action["thrust"]
 	var rot: Array = action["rotation"]
-	rl_controller.action.thrust = Vector3(thrust[0], thrust[1], thrust[2])
-	rl_controller.action.rotation = Vector3(rot[0], rot[1], rot[2])
+	var thrust_y: float = thrust[1] if allow_vertical else 0.0
+	var pitch: float = rot[0] if allow_pitch_roll else 0.0
+	var roll: float = rot[2] if allow_pitch_roll else 0.0
+	rl_controller.action.thrust = Vector3(thrust[0], thrust_y, thrust[2])
+	rl_controller.action.rotation = Vector3(pitch, rot[1], roll)
 	rl_controller.action.turbo = int(action["turbo"]) == 1