You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When a cartesian constraint goal is sent to STOMP via MoveIt, the target_frame is equal to world (see this line) and thus fails because the world frame is not part of the robot state. This behavior occurs with the panda_moveit_config from moveit_resources package because it uses a floating virtual joint. When a virtual joint is not a fixed frame, the model frame becomes the parent of the virtual joint.
If I change this line with this below it works normally:
Are you sure this also works when the virtual joint actually moves (and thus world and root link do not coincide anymore)? The IKSolver has additional logic for a tf_base_to_root_ transform, but I would not expect it to work as expected without testing.
Btw. looks like the logic here actually relies on frame information from a default-initializedRobotState which is not sufficient for constraints specified relative to the current pose and I would expect robot_state_ here instead. As I don't use stomp in any of my setups at the moment, I can't verify this though.
When a cartesian constraint goal is sent to STOMP via MoveIt, the
target_frame
is equal toworld
(see this line) and thus fails because the world frame is not part of the robot state. This behavior occurs with thepanda_moveit_config
frommoveit_resources
package because it uses a floating virtual joint. When a virtual joint is not a fixed frame, the model frame becomes the parent of the virtual joint.If I change this line with this below it works normally:
The text was updated successfully, but these errors were encountered: