My new CNC project HEXAPOD ROBOT
Posted by
KM6VV
on 2005-11-05 09:55:38 UTC
Hi to the lists,
My apologies for the multiple posts, but I think this may be a "crossover"
project of interest to members of several groups.
The initial HEXAPOD robot I want to build is basically a 12 degrees of freedom
CNC! OK, maybe that's an exaggeration, but I'm thinking I can "migrate" my
controller code to generate moves for the legs of a hexapod robot. I'm
considering using two or three steppers to "model" a leg, and allow me to test
motion directly with Gcode. Then I would substitute R/C servos for the steppers
in order to build the robot. But there is a big difference in size (and weight)
between a "standard" R/C servo and a stepper!
For each leg, we cycle the vertical and horizontal servos, each servo moving
between set angles. The servos take PWM signals, with a pulse varying between
0.5 - 2.5 ms, for 180 degrees. So the resolution is .09 degrees/step. The
pulse repeats about every 20 ms. So by generating a PWM signal instead of
steps, I can directly relate moves, and then write Gcode for them. Coordinated
moves are a cinch, (just a LOT of 'em)!
The 6 legs must move in coordinated motions called gaits. The most common gaits
are tripod and tetrapod, but there are others. This is part of what I want to
explore. So there is considerable coordination between legs, but each leg (two
or three servos) works much the same as any other leg; until it comes time to
change the horizontal distance to be covered, or lift the leg differently (when
trying to overcome obstacles).
I will of course be using Vector CAD/CAM and CNC on my Sherline mill to cut out
sheet metal parts. Not much of a CNC challenge, but VERY convenient! All of
the parts must be light, as there is barely enough power to move the robot. Ill
be able to do some limited 3D modeling as well (not as much as I'd like). If
all comes out the way I want, I'll make up plans to be available on my website
(do hexapod robots go with steam engines?).
Original "ServoBot":
http://www.cs.indiana.edu/~sjohnson/projects/servobot/
It used lexan plastic parts (LynxMotion made a commercial version). I'm
thinking thin aircraft grade aluminum would be lighter and stronger. I can
probably mill out several parts at once (stacked), and of course the drilling
will be made easy. Then there's the bending. Anybody got ideas there? I don't
have a brake, I've clamped sheet metal with machinist's clamps and bent them
before. Or maybe I can borrow a brake.
I NEED SOURCES of small sheets of aircraft aluminum (what's the grade?) I'd
also like to find some LEXAN to play with.
I have a designed a USB/RS232 18F4550 PIC to control the legs, and am "porting"
code from the SSC32 "open source" servo controller:
http://wibots.com/
I have a few changes in mind, 'tho. The current hexapod is two-servo-per-leg
and 6 legs, I'm thinking three-servo-per-leg and 8 legs, an "octabot", if you will.
Other good sites:
http://www.tarry.de/index_us.html
http://www.uni-bielefeld.de/biologie/Kybernetik/research/walk_legcoord.html
A few papers too:
http://cs.conncoll.edu/Parker/_papers.html
I probably need to find a list for hexapods! But not just a "kit builder" list,
I'm more into the design of the hardware/mechanics, and electronics/firmware.
Then there's the kinematics of leg movement.
Alan KM6VV
My apologies for the multiple posts, but I think this may be a "crossover"
project of interest to members of several groups.
The initial HEXAPOD robot I want to build is basically a 12 degrees of freedom
CNC! OK, maybe that's an exaggeration, but I'm thinking I can "migrate" my
controller code to generate moves for the legs of a hexapod robot. I'm
considering using two or three steppers to "model" a leg, and allow me to test
motion directly with Gcode. Then I would substitute R/C servos for the steppers
in order to build the robot. But there is a big difference in size (and weight)
between a "standard" R/C servo and a stepper!
For each leg, we cycle the vertical and horizontal servos, each servo moving
between set angles. The servos take PWM signals, with a pulse varying between
0.5 - 2.5 ms, for 180 degrees. So the resolution is .09 degrees/step. The
pulse repeats about every 20 ms. So by generating a PWM signal instead of
steps, I can directly relate moves, and then write Gcode for them. Coordinated
moves are a cinch, (just a LOT of 'em)!
The 6 legs must move in coordinated motions called gaits. The most common gaits
are tripod and tetrapod, but there are others. This is part of what I want to
explore. So there is considerable coordination between legs, but each leg (two
or three servos) works much the same as any other leg; until it comes time to
change the horizontal distance to be covered, or lift the leg differently (when
trying to overcome obstacles).
I will of course be using Vector CAD/CAM and CNC on my Sherline mill to cut out
sheet metal parts. Not much of a CNC challenge, but VERY convenient! All of
the parts must be light, as there is barely enough power to move the robot. Ill
be able to do some limited 3D modeling as well (not as much as I'd like). If
all comes out the way I want, I'll make up plans to be available on my website
(do hexapod robots go with steam engines?).
Original "ServoBot":
http://www.cs.indiana.edu/~sjohnson/projects/servobot/
It used lexan plastic parts (LynxMotion made a commercial version). I'm
thinking thin aircraft grade aluminum would be lighter and stronger. I can
probably mill out several parts at once (stacked), and of course the drilling
will be made easy. Then there's the bending. Anybody got ideas there? I don't
have a brake, I've clamped sheet metal with machinist's clamps and bent them
before. Or maybe I can borrow a brake.
I NEED SOURCES of small sheets of aircraft aluminum (what's the grade?) I'd
also like to find some LEXAN to play with.
I have a designed a USB/RS232 18F4550 PIC to control the legs, and am "porting"
code from the SSC32 "open source" servo controller:
http://wibots.com/
I have a few changes in mind, 'tho. The current hexapod is two-servo-per-leg
and 6 legs, I'm thinking three-servo-per-leg and 8 legs, an "octabot", if you will.
Other good sites:
http://www.tarry.de/index_us.html
http://www.uni-bielefeld.de/biologie/Kybernetik/research/walk_legcoord.html
A few papers too:
http://cs.conncoll.edu/Parker/_papers.html
I probably need to find a list for hexapods! But not just a "kit builder" list,
I'm more into the design of the hardware/mechanics, and electronics/firmware.
Then there's the kinematics of leg movement.
Alan KM6VV