The demo launch files started IMU nodes that were
unused, produced warnings, and created problematic tf trees.
With this commit, they no longer start these nodes.
Quick tests of some of the demos indicate that they
work as well as they did before. However, I did not test
hello_world.launch.
NOT FULLY TESTED
After these very minor changes, the following command
appeared to execute properly. However, I didn't bother
attaching the dry-erase marker tool. Instead, I placed
the robot in front of a whiteboard and caught and followed
the gripper with my hand at the surface of the board
to simulate lower-friction dry-erase marker contact.
It successfully aligned itself with the board, stopped on
contact with my hand, and completed all of the writing
motions.
$ roslaunch stretch_demos hello_world.launch
After these very minor changes, surface cleaning with the
following command worked for me. Stretch wiped the top of
a small nightstand.
$ roslaunch stretch_demos clean_surface.launch
After these changes, object grasping with the following command
worked for me. Stretch grasped a cup from a small nightstand.
$ roslaunch stretch_demos grasp_object.launch
Changes include the following:
+ / => // for some places where an integer output is required
+ changes for scikit_image (import skimage) 0.18.1
+ skimage.measure.label argument change
+ skimage.measure.regionprops argument change
After this minor change, drawer opening worked with the
following command.
$ roslaunch stretch_demos open_drawer.launch
The following persistent warning would be good to resolve.
[ WARN] [1622728896.394958823]: Could not obtain transform from imu_mobile_base to base_link. Error was Could not find a connection between 'base_link' and 'imu_mobile_base' because they are not part of the same tree.Tf has two or more unconnected trees.
With these updates, the handover_object.launch demo succeeded.
The robot successfully moved its hand to me three times in a
row when I moved myself to different positions. The handoff
poses seemed good.
A variety of changes were required to make this work, including
the following:
+ tab vs. spaces cleanup for funmap using "autopep8 -i funmap"
+ updates to cython compilation for Python 3 and recent cython
+ changed import statement pattern to "stretch_funmap.*"
Prior to my successful run of this demo, I encountered an error
related to get_robot_floor_pose_xya in HelloNode in hello_misc.py
(details below). It seems like it may have been a failed transform
lookup by get_p1_to_p2_matrix, which queries TF2 to obtain the
current estimated transformation from the robot's base_link frame
to the frame.
I did not make changes that I would expect to resolve this issue.
I only added a few print statements to hello_misc.py that I
intended to use for debugging. I have left the print statements
in place.
[ERROR] [1622656837.765018]: Error processing request: matmul: Input operand 0 does not have enough dimensions (has 0, gufunc core with signature (n?,k),(k,m?)->(n?,m?) requires 1)
['Traceback (most recent call last):\n', ' File "/opt/ros/noetic/lib/python3/dist-packages/rospy/impl/tcpros_service.py", line 632, in _handle_request\n response = convert_return_to_response(self.handler(request), self.response_class)\n', ' File "/home/ck/catkin_ws/src/stretch_ros/stretch_demos/nodes/handover_object", line 178, in trigger_handover_object_callback\n at_goal = self.move_base.forward(self.mobile_base_forward_m, detect_obstacles=False, tolerance_distance_m=tolerance_distance_m)\n', ' File "/home/ck/catkin_ws/src/stretch_ros/stretch_funmap/src/stretch_funmap/navigate.py", line 245, in forward\n xya, timestamp = self.node.get_robot_floor_pose_xya()\n', ' File "/home/ck/catkin_ws/src/stretch_ros/hello_helpers/src/hello_helpers/hello_misc.py", line 127, in get_robot_floor_pose_xya\n r0 = np.matmul(robot_to_odom_mat, r0)[:2]\n', 'ValueError: matmul: Input operand 0 does not have enough dimensions (has 0, gufunc core with signature (n?,k),(k,m?)->(n?,m?) requires 1)\n']
Python 3 was not well supported by ROS Melodic.
stretch_deep_perception required Python 3, which could
cause issues and confusion. To help clarify when code
used Python 3, the filenames had the pattern *_python3.py.
I've changed the filenames to remove this pattern,
since all of the ROS Noetic code will be using Python 3
by default, and ROS Noetic supports Python 3.
+ move up or down to contact handles
+ use new stretch_driver custom effort thresholds
+ successful with tool chest drawer, but has 10 second timeouts after each move until contact