Subject:
|
Designing a bipedal robot leg
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Sun, 19 May 2002 02:17:31 GMT
|
Original-From:
|
Andy Gombos <gombos_2000@earthlink.&IHateSpam&net>
|
Viewed:
|
538 times
|
| |
|
|
Around a year ago I suppose I asked about curved feet - building a robot
with human-like walking characteristics. I never succeeded in the feet
(although it turns out some carved foam works very well.), as the robot took
one complete step but could never balance quickly enough to return to the
standing state.
*sigh*
After watching some robot show on the Discovery Science channel, I decided
to improve on the old leg design. However, days of searching have lead me
nowhere. I have made two designs:
-------------------------------------------------
-------------------------------------------------
\ /
\ /
\ / Piston
Piston \ /
\/
\ /
| \
/ |
Piston | \
/ | Piston
| /
\ |
/
\
/
\
/
\
/
\
-----------
-----------
The left one is a humanlike design, while the right one is like MITs Spring
Turkey or an AT-ST. The pneumatic system is too slow and unprecise, but Leo
Dorst's translational joint(motorized piston) is too large for the knee, and
makes balancing more difficult.
If you were to design a leg for a bipedal robot, how would you do it? Would
it have a knee, or a moveable ankle? Would you use pneumatics, motors and
cables, or the various muscle wires or pneumatic muscles available?
Another problem is balance. Balancing the previous design (which used the
legs on the right) required a full RCX (batteries and an extra motor) to be
extended over the side of the robot, past the plane of the leg.
----------------------
| |
| RCX |
---------------------------------------
LEG
---------------------------------------
Many times, the RCX fell off, became unbalanced, or didn't properly
counterbalance the legs. Would shortening the distance between the legs
help in this? It would seem to at first thought, but a larger weight would
be needed to cause a change in the forces on the moving leg. How would you
balance the robot on one leg while the other moved? Assume the leg will
take ~1 second to move to a position where it can support weight again, as
the knee must pull up, the leg is pivoted forward, the knee is re-extended,
and the counterbalance is tipped to allow weight to be placed onto the new
leg.
I feel like this should be fairly easy to produce a stable motion, even a
stable standing cycle, but it has proven impossible. I could do a
potentially simpler design that, while it would work, would not meet my
original goal.
It appears all who said this idea was fairly impossible with Lego was
correct - the scale and power of Legos are too small for proven designs to
be applied.
Andy
|
|
Message has 1 Reply:
 | | Re: Designing a bipedal robot leg
|
| Hi Andy, 2 links for ya: (URL) - my collection of links to Lego Bipeds on the web. Although it seems many technicques have been tried succesfully, there are new ways coming out constantly. I try to keep up. For bipedal legs, try Leo's Lego: (URL) (22 years ago, 21-May-02, to lugnet.robotics)
|
3 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
