diff --git a/flybody/fly_envs.py b/flybody/fly_envs.py
index 1ee347d..c5d5881 100755
--- a/flybody/fly_envs.py
+++ b/flybody/fly_envs.py
@@ -15,8 +15,11 @@
 from flybody.tasks.arenas.ball import BallFloor
 from flybody.tasks.arenas.hills import SineBumps, SineTrench
 from flybody.tasks.pattern_generators import WingBeatPatternGenerator
-from flybody.tasks.trajectory_loaders import (HDF5FlightTrajectoryLoader,
-                                              HDF5WalkingTrajectoryLoader)
+from flybody.tasks.trajectory_loaders import (
+    HDF5FlightTrajectoryLoader,
+    HDF5WalkingTrajectoryLoader,
+    InferenceWalkingTrajectoryLoader,
+)
 
 
 def flight_imitation(wpg_pattern_path: str,
@@ -60,13 +63,15 @@ def flight_imitation(wpg_pattern_path: str,
                                 strip_singleton_obs_buffer_dim=True)
 
 
-def walk_imitation(ref_path: str,
+def walk_imitation(ref_path: str | None = None,
                    random_state: np.random.RandomState | None = None,
                    terminal_com_dist: float = 0.3):
     """Requires a fruitfly to track a reference walking fly.
 
     Args:
-        ref_path: Path to reference trajectory dataset.
+        ref_path: Path to reference trajectory dataset. If not provided, task
+            will run in inference mode with InferenceWalkingTrajectoryLoader,
+            without loading actual walking dataset.
         random_state: Random state for reproducibility.
         terminal_com_dist: Episode will be terminated when distance from model
             CoM to ghost CoM exceeds terminal_com_dist. Can be float('inf').
@@ -77,8 +82,13 @@ def walk_imitation(ref_path: str,
     walker = fruitfly.FruitFly
     arena = floors.Floor()
     # Initialize a walking trajectory loader.
-    traj_generator = HDF5WalkingTrajectoryLoader(
-        path=ref_path, random_state=random_state)
+    if ref_path is not None:
+        inference_mode = False
+        traj_generator = HDF5WalkingTrajectoryLoader(
+            path=ref_path, random_state=random_state)
+    else:
+        inference_mode = True
+        traj_generator = InferenceWalkingTrajectoryLoader()
     # Build a task that rewards the agent for tracking a walking ghost.
     time_limit = 10.0
     task = WalkImitation(walker=walker,
@@ -87,6 +97,7 @@ def walk_imitation(ref_path: str,
                          terminal_com_dist=terminal_com_dist,
                          mocap_joint_names=traj_generator.get_joint_names(),
                          mocap_site_names=traj_generator.get_site_names(),
+                         inference_mode=inference_mode,
                          joint_filter=0.01,
                          future_steps=64,
                          time_limit=time_limit)