diff --git a/lerobot/common/envs/utils.py b/lerobot/common/envs/utils.py
index 78517abe..6003acc0 100644
--- a/lerobot/common/envs/utils.py
+++ b/lerobot/common/envs/utils.py
@@ -47,7 +47,9 @@ def preprocess_observation(observations: dict[str, np.ndarray]) -> dict[str, Ten
             # TODO(aliberts, rcadene): use transforms.ToTensor()?
             img = torch.from_numpy(img)
 
-            if img.dim() == 3:
+            # When preprocessing observations in a non-vectorized environment, we need to add a batch dimension.
+            # This is the case for human-in-the-loop RL where there is only one environment.
+            if img.ndim == 3:
                 img = img.unsqueeze(0)
             # sanity check that images are channel last
             _, h, w, c = img.shape
diff --git a/lerobot/scripts/server/kinematics.py b/lerobot/scripts/server/kinematics.py
index d6b712c8..dc5baf54 100644
--- a/lerobot/scripts/server/kinematics.py
+++ b/lerobot/scripts/server/kinematics.py
@@ -51,6 +51,15 @@ def screw_axis_to_transform(S, theta):
 def pose_difference_se3(pose1, pose2):
     """
     Calculates the SE(3) difference between two 4x4 homogeneous transformation matrices.
+    SE(3) (Special Euclidean Group) represents rigid body transformations in 3D space, combining rotation (SO(3)) and translation.
+    Each 4x4 matrix has the following structure, a 3x3 rotation matrix in the top-left and a 3x1 translation vector in the top-right:
+    
+    [R11 R12 R13 tx]
+    [R21 R22 R23 ty]
+    [R31 R32 R33 tz]
+    [ 0   0   0   1]
+    
+    where Rij is the 3x3 rotation matrix and [tx,ty,tz] is the translation vector.
 
     pose1 - pose2