Trace:

Differences

This shows you the differences between two versions of the page.

--- projects:cucumber_cutting [2012/06/29 18:52] – added heading kresse
+++ projects:cucumber_cutting [2012/07/10 14:48] (current) – corrected coordinate system description kresse
@@ Line 1: / Line 1: @@
+====== Cucumber Cutting, magnetically tracked ======
+This page hosts some experimental data from cucumber cutting which was recorded
+with the magnetic motion tracker 'Razor Hydra (tm)'.
+The following images show the setup before the start of the experiment.
+{{:projects:cucumber_cutting:cucumber_cutting_setup.jpg?direct&300|cucumber cutting setup}}
+{{:projects:cucumber_cutting:cucumber_cutting_setup_top.jpg?direct&300|cucumber cutting setup (top)}}
+{{:projects:cucumber_cutting:cucumber_cutting_setup_side.jpg?direct&300|cucumber cutting setup (side)}}
+{{:projects:cucumber_cutting:cucumber_cutting_setup_front.jpg?direct&300|cucumber cutting setup (front)}}
+Then, the cucumber was cut into slices, starting from the right end (phase1). after about cutting half of the cucumber, the knife was put aside and the cucumber was moved:
+{{:projects:cucumber_cutting:cucumber_cutting_setup_phase2.jpg?direct&300|cucumber cutting setup (before second phase)}}
+Then the cutting continued (phase2) until all of the cucumber was cut.
+**Beware**: This may be a challenging dataset due to systematic errors in the
+tracking device. Hopefully, these errors are small in the relevant workspace.
+===== Coordinate System =====
+The origin is in the green ball (the emitter), with z pointing down, y pointing to the left and x  pointing towards the cutting person.
+===== Calibration =====
+I tried to identify points of interest, both on the tool and in the scene.
+Until the calibration code for that is polished, I post the calibration points here in text form.
+tip-of-blade:
+  [0.29293112, 0.03024254, 0.03524071]
+grip-end-of-blade edge:
+  [[ 0.16761572] [ 0.00385015] [ 0.04192011]]
+normal of the blade:
+pose of the paddle, when the blade is lying flat on the board:
+  [[    0.993666,    0.073076,   0.0853739;
+      -0.105184,    0.337344,    0.935487;
+.0395613,   -0.938541,    0.342893]
+  [       -0.62,       0.164,       0.144]]
+lying on the other side (here, the mounting of the paddle was bending noticably):
+  [[    0.948797,    0.313658,   0.0374525;
+.160745,   -0.377344,   -0.912016;
+      -0.271929,    0.871338,   -0.408442]
+  [      -0.621,       0.162,       0.225]]
+table corners:
+  [   -0.479589,     -0.3937,    0.204708]
+  [   -0.375432,    0.196603,    0.174091]
+  [third corner is nonsense]
+cutting board (highly distorted):
+  [   -0.618805,   -0.127104,   0.0772075]
+  [   -0.475962,    0.179408,   0.0621018]
+  [   -0.174584,    0.142956,   0.0148497]
+===== Recording =====
+The cucumber ends on the board were as follows:
+cucumber ends:
+  [   -0.504385,   -0.126844,    0.226978]
+  [   -0.422555,    0.285663, -0.00650265]
+Then, I did two recordings of cutting (phase1), then I moved the cucumber:
+cucumber ends (second round):
+  [   -0.410377,   -0.109242,    0.311248]
+  [   -0.470199,    0.136546,    0.143661]
+Then the recording phase2 happened.
+===== References =====
+  * The ros bags with the recordings are {{:projects:cucumber_cutting:cucumber_cutting_2012-06-29.tar.bz2|here}}.
+  * The cutting itself is also avaliable as gnuplot/matlab compatible text files {{:projects:cucumber_cutting:cutting_logs.tar.bz2|here}}.
+  * The data are transformed from the Hydra message format to a ''geometry_msgs/TransformStamped'' message by the node ''hydra_tf_node''.
+  * The code is in my sandbox in ''hydra_tf_node'' and ''calib_tools'':
+  hg clone http://toychest.in.tum.de/users/kresse/sandbox