Fixes bugs in math operation in `omni.isaac.orbit.utils.math` (#55)

* dealing with negative w component in quat
* fix bug on subtract_fram_transform by applying the rotation
* update version and changelog

source/extensions/omni.isaac.orbit/config/extension.toml

 [package]
 
 # Note: Semantic Versioning is used: https://semver.org/
-version = "0.2.7"
+version = "0.2.8"
 
 # Description
 title = "ORBIT framework for Robot Learning"

source/extensions/omni.isaac.orbit/docs/CHANGELOG.rst

 Changelog
 ---------
 
+0.2.8 (2023-04-10)
+~~~~~~~~~~~~~~~~~~
+
+Fixed
+^^^^^
+
+* Fixed bugs in :meth:`axis_angle_from_quat` in the ``omni.isaac.orbit.core.utils.math`` to handle quaternion with negative w component.
+* Fixed bugs in :meth:`subtract_frame_transforms` in the ``omni.isaac.orbit.core.utils.math`` by adding the missing final rotation.
+
+
 0.2.7 (2023-04-07)
 
 Fixed

source/extensions/omni.isaac.orbit/omni/isaac/orbit/utils/math.py

@@ -329,7 +329,7 @@ def axis_angle_from_quat(quat: torch.Tensor, eps: float = 1.0e-6) -> torch.Tenso
     # Thus, axis-angle is [q_x, q_y, q_z] / (sin(theta/2) / theta)
     # When theta = 0, (sin(theta/2) / theta) is undefined
     # However, as theta --> 0, we can use the Taylor approximation 1/2 - theta^2 / 48
-
+    quat = quat * (1.0 - 2. * (quat[..., 0:1] < 0.0))
     mag = torch.linalg.norm(quat[..., 1:], dim=1)
     half_angle = torch.atan2(mag, quat[..., 0])
     angle = 2.0 * half_angle
@@ -398,9 +398,10 @@ def subtract_frame_transforms(
             frame 2 w.r.t. frame 1.
     """
     # compute orientation
-    q12 = quat_mul(quat_inv(q01), q02)
+    q10 = quat_inv(q01)
+    q12 = quat_mul(q10, q02)
     # compute translation
-    t12 = t02 - t01
+    t12 = quat_apply(q10, t02 - t01)
 
     return t12, q12