ROS and spatial algebra semantics

Last Edited: September 1, 2022 2:32 PM

When looking up transformation on ROS tf, keep in mind that if you want a transformation \(^{A}X^{B}\), (read as transformation from the frame \(A\) to frame \(B\)), then the way you would articulate the lookup in lookup_transform function in tf would be

buf = tf2_ros.Buffer()
tf_listener = tf2_ros.TransformListener(buf)
transform = buf.lookup_transform(A, B, rospy.Time(0))

Don’t let the target to source terminology confuse you here. More about this. The above function returns the transformation \(^{A}X^{B}\).

In Tedrake’s Robotic Manipulation, the usual notation has a target frame (say G), reference frame(say F) and a expressed in frame (say F again). The expressed in frame is dropped when talking about pure rotations and transformations, \(R, X\).

\[^{F}X^{ G}, ^{F}R^{ G}, ^{F}p^{ G}_{ F}\]

In ROS, the main message used to depict transformations is the TransformStamped message. This has attributes frame_id and child_frame_id. Then in Tedrake’s notation the transformation extracted from the above ros functionality is as follows.

\[^{frame\_id}X^{ child\_frame\_id}\]