Backhoe digger

[From Richard Kennaway (2004.01.14.2200 GMT)]

I've now made a smaller version of the digger movie, linked from
the same commentary page
http://www.cmp.uea.ac.uk/~jrk/Robotics/digger/digger.html. It's 1MB
in QuickTime format, and about the size of a large postage stamp.

-- Richard Kennaway

[From Bill Williams 14 January 2OO4 7:00 PM CST]

Couldn't view even the smaller digger movie-- our university IS people have the system tied up so it is imposible to do much of anything-- I guess that is their reference level for sucess.

I would think your work might have very widespread application. Sure would make it easier for a naive user to control a backhoe.

But, fancinating stuff just for its own sake.

Bill Williams

[From Bill Powers (2004.01.14..1849 MST)]

Richard Kennaway (2004.01.14.2200 GMT)--

Excellent! I had to download a newer version of Quicktime (11 MB!) but the
result was worth it.

I suggest working out a manual control like something a real operator of
the machine would use. A horizontal bar mounted crosswise (like bicycle
handlebars) on a linkage in front of the operator could be grasped, and
used to move the cutting edge of the scoop. Twisting the bar by rotating it
forward and back would tilt the scoop. The scoop would imitate the
operator's movements. The other hand could be used to select preset
directions with buttons, the way the control key is used in a Paint program
to make drawn lines exactly horizontal or vertical.

Also, I suggest auxiliary legs with their feet placed lightly on the ground
to define a reference position and reference plane, jointed to the body of
the digger through angle sensors. When the digger encounters heavy
resistance, its body is likely to move relative to the ground, so the
position of the scoop should be measured relative to a stable frame of
reference. When precision is needed, as in avoiding buried objects, the
auxiliary legs can provide a stable coordinate system.

This is further application of the principle of using sensors rather than
brute strength and mass to achieve a desired result.

I am reminded of the Canadian robot arms on the Space Shuttle and ISS. The
controls can, I believe, be used to move the arm in desired spatial
directions rather than having the degrees of freedom individually
controlled. But I have wondered how they achieve stability of the end of
the arm. Is it done by precalculating torque patterns applied to the
joints, or is there some means of sensing the actual position of the
end-effector, say with laser beams inside the tubes of the arm segments? If
actual positions and angles of the load could be sensed, they could be
controlled by applying forces and torques as if the load were part of the
arm, and springiness or bending of the arm would have no effect.

What do you think of the idea of a Mars Walker?

Best,

Bill P.

[From Rick Marken (2004.01.16.1040)]

Richard Kennaway (2004.01.14.2200 GMT)

I've now made a smaller version of the digger movie, linked from
the same commentary page.
Robotics and control theory simulations.
It's 1MB in QuickTime format, and about the size of a large postage stamp.

This is a really nifty demo, Richard. I'm going to point to it from my site.
Do you have a figure that describes the model? That would be a nice
supplement to the Commentary. I'd like to see a program listing, too. Is
that something you'd be willing to post.

You do really great PCT modeling, Richard. Thanks!

Best

Rick

···

---
Richard S. Marken
MindReadings.com
Home: 310 474 0313
Cell: 310 729 1400

From Richard Kennaway (2004.01.16.1846 GMT):

[From Rick Marken (2004.01.16.1040)]

Richard Kennaway (2004.01.14.2200 GMT)

I've now made a smaller version of the digger movie, linked from
the same commentary page.

>Robotics and control theory simulations.

  It's 1MB in QuickTime format, and about the size of a large postage stamp.

This is a really nifty demo, Richard. I'm going to point to it from my site.
Do you have a figure that describes the model? That would be a nice
supplement to the Commentary. I'd like to see a program listing, too. Is
that something you'd be willing to post.

No figure yet, this is hot off the software development system. I
intend to put this, the robot, a few other mechanical simulations,
and some mathematical analysis into a paper.

A colleague suggested making the task more difficult by adding extra
joints to the arm. Notice that a backhoe has three joints (ignoring
turning from side to side), and there are also three
operator-controlled variables: shovel height, depth, and orientation.
Adding a fourth joint introduces an extra degree of freedom into the
system that the operator isn't controlling.

                       2-------3
                      / \
                     / \
                    / \
                   / ----4
                  /
         +-----+ /

···

    >/

         > 1
         +-----+

If you imagine the base and the shovel held fixed, you would be able
to exercise the extra degree of freedom in the arm linkage without
being resisted by the controllers.

So to keep this machine stable, one has to find another variable to
control. I decided that keeping consecutive joint angles (excluding
the first and last joints) equal seemed a reasonable thing to do. In
the above picture that would be joint angles 2 and 3.

I've put a couple of movies of the result, for a 5-jointed digger, in
Robotics and control theory simulations, files
digger5joints.avi and dig5eq.avi. Again, these are rather large AVI
files. I'll convert them to smaller QuickTime movies when I get home
tonight, I don't have the software at work.

digger5joints.avi shows a 5-jointed digger without the angle
equalising controllers, and dig5eq.avi shows it with. The equaliser
controllers actually control a(i+1) - k*a(i) rather than a(i+1) -
a(i), where k is 0.9. I picked that formula by trial and error,
finding that some formulas gave instabilities.

--
Richard Kennaway, jrk@cmp.uea.ac.uk, Richard Kennaway
School of Computing Sciences,
University of East Anglia, Norwich NR4 7TJ, U.K.

[From Rick Marken (2004.01.16.1630)]

Richard Kennaway (2004.01.16.1846 GMT)

A colleague suggested making the task more difficult by adding extra
joints to the arm...

                     2-------3
                    / \
                   / \
                  / \
                 / ----4
                /
       +-----+ /
       > >/
       > 1
       +-----+

I've put a couple of movies of the result, for a 5-jointed digger, in
Robotics and control theory simulations, files
digger5joints.avi and dig5eq.avi.

The movies are great!! This stuff is just too cool!! I love the little
disturbances that have no effect on the tilt of the hand.

Wonderful work, Richard.

Best

Rick

···

---
Richard S. Marken
MindReadings.com
Home: 310 474 0313
Cell: 310 729 1400