Differences

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

--- ece4560:lynx6:calibration [2017/10/22 16:42] – created pvela
+++ ece4560:lynx6:calibration [2024/08/20 21:38] (current) – external edit 127.0.0.1
@@ Line 4: / Line 4: @@
 This module deals with calibration of the manipulator in the straight-up home configuration (basically the manipulator arm points straight up into the air).  A straight up home configuration means that this is the zero configuration for the robot.  All joint angles are measured from this
 reference configuration. Calibration will require measuring the link lengths associated to your manipulator, the servo command limits for each of the joints, and the angular limits of the manipulator.  The limits imposed may actually be less than the full range of motion of the servo motors.  The sub-problems below lead you through the procedure.
-. Measure the lengths of all the links.  Make sure to include
-    the wooden base in your measurements for vertical height as appropriate.
-    Try to be as accurate as possible since mistakes here will impact your
-    ability to do useful things with the manipulator.
-    There is a member variable in the {\tt lynx6} class object for
+  - Measure the lengths of all the links.  Make sure to include the base in your measurements for vertical height as appropriate. Try to be as accurate as possible since mistakes here will impact your ability to do useful things with the manipulator. \\ There is a member variable in the {\tt lynx6} class object for specifying these variables.  They've been set based on my setup, but may be different for yours.  The numbers in the list are the base to shoulder height, the length of the first link, the length of the second link, then the link to middle of gripper length, and the additional length to gripper tip.
-    specifying these variables.  They've been set based on my setup,
+  - Determine what are the proper servo command limits for your particular manipulator.  The manipulator code for interfacing is located [[ECE4560:ManiControl | here]] with links describing in more detail what calibration entails. Please read the document twice and be careful, as the manipulators break easily.
-    but may be different for yours.  The numbers in the list are
+  - The last phase is to associate angles to the servo commands. Using the straight-up reference configuration as the zero configuration for the joint-angles, determine the angular limits associated to the joints with respect to that reference configuration as best as you can.  You *will* need a protractor. \\  Although the robot may be able to go to bigger angles, try to round the joint-angle workspace to the nearest 30, 45, 90, or 180 degrees. For example, if you measure that it can go -37 degrees to 48 degrees, then make the joint angle limits $[-30, 45]$ in order to have nice clean limits.  You'll need to figure out what servo commands correspond to those particular angle limits.
-    the base to should height, the length of the first link,
-    the length of the second link, then the link to middle of gripper
-    length, and the additional length to gripper tip.
-. Dtermine what are the proper servo command limits for your
-    particular manipulator.  The manipulator code for interfacing should be
-    uploaded to {\tt t-square} as will a file describing in more detail
-    what this problem is about.  Please read the document twice and be
-    careful, they break easily.
-. The last phase is to associate angles to the servo commands.
+Save it into your own file whose default is called ''lynx6parms.mat'' for use with the ''lynx6'' class file.  Turn in the values of the structure in the ''lynx6parm'' matlab file as well as your link lengths.  Some folk have been turning in a screenshot of the calibration dialog window.  That works too (''Ctrl-PrintScreen'' will capture it and ''Ctrl-V'' will paste the image).  If you save the screen capture to an image file, you should even be able to load the image file, display it in a Matlab figure, then have it be output as part of the ''publish'' execution.
-    Using the straight-up reference configuration as the zero
-    configuration for the joint-angles, determine the angular
-    limits associated to the joints with respect to that reference
-    configuration as best as you can.  You WILL need a protractor.
-    Although the robot may be able to go to bigger angles, try to round
+==== Demo: Verification ====
-    the joint-angle workspace to the nearest 30, 45, 90, or 180 degrees.
-    For example, if you measure that it can go -37 degrees to 48 degrees,
-    then make the joint angle limits $[-30, 45]$ in order to have nice
-    clean limits.  You'll need to figure out what servo commands
-    correspond to those particular angle limits.
-Save it into your own file whose default is called {\tt lynx6parms.mat}
+To make sure that all the manipulators are calibrated, the demo will involve a verification exercise.  To that end, I (or the TA) will run some code on your manipulator to do the following:
-for use with the {\tt lynx6} class file.  Turn in the values of the
+  - Invoke ''setArm'' with all zeros to verify that your traight up really is straight up;
-structure in the {\tt lynx6parm} matlab file as well as your link
+  - Invoke ''setArm'' with the proper values to make it go straight out;
-lengths.
+  - Call a ''for'' loop that will go to 5 random configurations, to be judged (subjectively) for correctness; then finally
+  - Invoke the forward kinematics routine, that you will have coded, for a specific set of joint angles and go to the desired joint angles.  You will be graded based on how close your    manipulator is to the predicted forward kinematics.  This module part requires for you to code up the forward kinematics function in the ''lynx6'' m-file. The function should return an element of the ''SE3'' class.
-To make sure that all the manipulators are calibrated, the demo
+I will post to t-square some code to do the main loop that you should demo to I or the TA to test for correct calibration.  It is similar to the piktul code that was run and includes the predicted opening and closing widths of the gripper.
-will involve a verification exercise.  To that end, I (or the TA) will
-run some code on your manipulator to do the following:
-\begin{description}
-  \item[(a)]  Invoke {\tt setArm} with all zeros to verify that your
-    straight up really is straight up;
-  \item[(b)] Invoke {\tt setArm} with the proper values to make it go
-    straight out;
-  \item[(c)] Call a for loop that will go to 5 random configurations, to
-    be judged (subjectively) for correctness; then finally
-  \item[(d)] Invoke the forward kinematics routine, that you will have
-    coded, for a specific set of joint angles and go to the desired
-    joint angles.  You will be graded based on how close your
-    manipulator is to the predicted forward kinematics.  This problem
-    part requires for you to code up the forward kinematics function in
-    {\tt lynx6.m}.  The function should return an element of the
-    {\tt SE3} class.
-\end{description}
-I will most likely post some code to do the main loop that you should
-demo to I or the TA to test for correct calibration.  It is similar to
-the {\tt piktul}  code that was run and includes the predicted opening
-and closing widths of the gripper.
-%What you will bring to the table is the {\tt lynx6.m} file that is
-%specific to your manipulator.
-%You should also code up the
-%forward kinematics routine within that m-file.
-If you wonder what the base coordinate frame should be, just look at
-Figure \figref{man_lynx6_up}.  That is what I, and therefore you, will
-be using.  Code approriately.