[jsk_pcl_ros] I don't know how to use depth_error_calibration.py

kochigami commented 8 years ago

I'm following this tutorial in order to learn how to calibrate kinect sensor. (In the future, I'd like to try it on pepper.)

However, when I executed rosrun jsk_pcl_ros depth_error_calibration.py --model quadratic-uv-quadratic-abs, I got stuck.

My environment is ubuntu 14.04 and ros indigo. I use jsk_recognition from source. I changed <include file="$(find openni2_launch)/launch/openni2.launch"> to <include file="$(find freenect_launch)/launch/freenect.launch"> of openni2_remote(and _local) .launch . (I don't know why, but these programs didn't run for my kinect.)

I experienced these while showing checker board to kinect, press Ctrl+c and y button.

arrow, checker board appeared in rviz (while showing checker board)
failed in plotting graph by matplotlib (while showing checker board)
making calibration csv file, which content is null (after pressing Ctrl+c)
failed in making calibration yaml file (after pressing y)

Error message is as follows:

rosrun jsk_pcl_ros depth_error_calibration.py --model quadratic-uv-quadratic-abs
/home/kochigami/catkin_ws/src/jsk_recognition/jsk_pcl_ros/scripts/check_depth_error/depth_error_calibration.py:354: SyntaxWarning: The publisher should be created with an explicit keyword argument 'queue_size'. Please see http://wiki.ros.org/rospy/Overview/Publishers%20and%20Subscribers for more information.
^CSave calibration parameters to calibration-2016-05-18-14-44-09.csv
Dump result into yaml file? [y/n] Exception in Tkinter callback
Traceback (most recent call last):
  File "/usr/lib/python2.7/lib-tk/Tkinter.py", line 1489, in __call__
    return self.func(*args)
  File "/usr/lib/python2.7/lib-tk/Tkinter.py", line 536, in callit
    func(*args)
  File "/usr/lib/pymodules/python2.7/matplotlib/backends/backend_tkagg.py", line 363, in idle_draw
    self.draw()
  File "/usr/lib/pymodules/python2.7/matplotlib/backends/backend_tkagg.py", line 349, in draw
    tkagg.blit(self._tkphoto, self.renderer._renderer, colormode=2)
  File "/usr/lib/pymodules/python2.7/matplotlib/backends/tkagg.py", line 20, in blit
    tk.call("PyAggImagePhoto", photoimage, id(aggimage), colormode, id(bbox_array))
TclError: this isn't a Tk application
y
writing to calibration_parameter_2016_05_18_14_44_12.yaml
Traceback (most recent call last):
  File "/home/kochigami/catkin_ws/src/jsk_recognition/jsk_pcl_ros/scripts/check_depth_error/depth_error_calibration.py", line 434, in <module>
    main()
  File "/home/kochigami/catkin_ws/src/jsk_recognition/jsk_pcl_ros/scripts/check_depth_error/depth_error_calibration.py", line 414, in main
    c = classifier.coef_
AttributeError: 'LinearRegression' object has no attribute 'coef_'

I think

Failed in showing graph by matplotlib relates to fail in using matplotlib on linux
Attribute error relates to sklearn package version

sklearn path is below

Python 2.7.6 (default, Jun 22 2015, 17:58:13) 
[GCC 4.8.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import sklearn
>>> sklearn.__path__
['/usr/lib/python2.7/dist-packages/sklearn']

but I'm not sure. Could anyone help me? Sorry for my trouble.

wkentaro commented 8 years ago

Traceback (most recent call last): File "/home/kochigami/catkin_ws/src/jsk_recognition/jsk_pcl_ros/scripts/check_depth_error/depth_error_calibration.py", line 434, in main() File "/home/kochigami/catkin_ws/src/jsk_recognition/jsk_pcl_ros/scripts/check_depth_error/depth_errorcalibration.py", line 414, in main c = classifier.coef AttributeError: 'LinearRegression' object has no attribute 'coef_'

The error seems to be caused by not passed --csv option. https://github.com/jsk-ros-pkg/jsk_recognition/blob/master/jsk_pcl_ros/scripts/check_depth_error/depth_error_calibration.py#L388-L401

So, I think the upper error (below) is the actual cause.

rosrun jsk_pcl_ros depth_error_calibration.py --model quadratic-uv-quadratic-abs /home/kochigami/catkin_ws/src/jsk_recognition/jsk_pcl_ros/scripts/check_depth_error/depth_error_calibration.py:354: SyntaxWarning: The publisher should be created with an explicit keyword argument 'queue_size'. Please see http://wiki.ros.org/rospy/Overview/Publishers%20and%20Subscribers for more information. ^CSave calibration parameters to calibration-2016-05-18-14-44-09.csv Dump result into yaml file? [y/n] Exception in Tkinter callback Traceback (most recent call last): File "/usr/lib/python2.7/lib-tk/Tkinter.py", line 1489, in call return self.func(_args) File "/usr/lib/python2.7/lib-tk/Tkinter.py", line 536, in callit func(_args) File "/usr/lib/pymodules/python2.7/matplotlib/backends/backend_tkagg.py", line 363, in idle_draw self.draw() File "/usr/lib/pymodules/python2.7/matplotlib/backends/backend_tkagg.py", line 349, in draw tkagg.blit(self._tkphoto, self.renderer._renderer, colormode=2) File "/usr/lib/pymodules/python2.7/matplotlib/backends/tkagg.py", line 20, in blit tk.call("PyAggImagePhoto", photoimage, id(aggimage), colormode, id(bbox_array)) TclError: this isn't a Tk application y

kochigami commented 8 years ago

Thank you very much for your help.

It may relate to my kinect version. Once I can find kinect one, I will try it with openni2_local (remote).launch. Thank you very much again.

What I found

unpublished depth_image_error/output topic from depth_error.launch I found in order to draw graph in this program, depth_image_error/output topic from jsk_pcl_ros/launch/depth_error.launch is required, but it is not published in my case. I also found my kinect is not kinect one, which is recommended to use in this tutorial. I'd like to find kinect one and try openni2_local(remote).launch first.
no problem in depth_error.launch I also compared how to use depth_image_error program in depth_error.launch, but it is no problem.

■ no change in depth_error.launch (use jsk_pcl_ros package) (I'm still uncertain why it works even though jsk_pcl_ros package doesn't seem to have depth_image_error.cpp or something..)

<node pkg="jsk_pcl_ros" type="depth_image_error"
        name="depth_image_error"
        output="screen">

■ use jsk_pcl_utils nodelet

 <node pkg="nodelet" type="nodelet"
        name="depth_image_error"
        args="standalone jsk_pcl_utils/DepthImageError"
        output="screen">

comparison of rosnode info /depth_image_error:

rosnode info /depth_image_error (jsk_pcl_ros package version)
--------------------------------------------------------------------------------
Node [/depth_image_error]
Publications: 
 * /depth_image_error/output [jsk_recognition_msgs/DepthErrorResult]
 * /rosout [rosgraph_msgs/Log]

Subscriptions: None

Services: 
 * /depth_image_error/list
 * /depth_image_error/load_nodelet
 * /depth_image_error/set_logger_level
 * /depth_image_error/get_loggers
 * /depth_image_error/unload_nodelet

contacting node http://192.168.97.25:44630/ ...
Pid: 13242
Connections:
 * topic: /rosout
    * to: /rosout
    * direction: outbound
    * transport: TCPROS

depth_error_image をnodeletで書いた時
 rosnode info /depth_image_error 
--------------------------------------------------------------------------------
Node [/depth_image_error]
Publications: 
 * /depth_image_error/output [jsk_recognition_msgs/DepthErrorResult]
 * /rosout [rosgraph_msgs/Log]

Subscriptions: None

Services: 
 * /depth_image_error/set_logger_level
 * /depth_image_error/get_loggers

contacting node http://192.168.97.25:55356/ ...
Pid: 6159
Connections:
 * topic: /rosout
    * to: /rosout
    * direction: outbound
    * transport: TCPROS

jsk-ros-pkg / jsk_recognition

[jsk_pcl_ros] I don't know how to use depth_error_calibration.py #1675