BasicMicro - Forums

Moved from RoboClaw topic.

Hi Nathan,

Just read your comments on Trig on the 'Pros and IK calcs.
Moved from AdamPro topic.

I've just bought a used Scorbot robot arm :

http://rssc.org/content/my-scorbot-robot-arm

And the simple control program an driver box don't appear to support coordinated moves. The arm uses geared 12 V. pitman DC servo motors with modest (6 step) encoders. Typical Jointed Arm Revolute construction. While I'm OK on leg and arm IKs, this arm has belts that pass over pulleys out to the distal end (grippers), so there is a parallelogram action, i.e., the grippers tend to stay in the same aspect relative to the arm (same plane). That's a little different from most LM (R/C Servo) arms! I'd think one would just "add in" the angle of the previous joint to the distal joints. Is that right? Have you written code for these?

I'm considering driving the arm directly with 4 RoboClaws and the Mad Hatter. That would be cool!

Alan KM6VV

_________________
Visit:
http://groups.yahoo.com/group/SherlineCNC/
http://tech.groups.yahoo.com/group/HexapodRobotIK/

Sorry, never saw one of those before. But I suspect you are correct. Either adding the angle or subtracting the angle to compensate for the parallelogram effect.

_________________
Tech Support
Basic Micro - Robotic Technology Evolved

Hi Alan,

Dunno if you still frequent this forum, but I'm looking to do a similar thing. I've got a Scorbot ER-III that I can control pretty well with some custom software and a USB->serial adapter with the stock controller, but the limited status updates, etc... are forcing me to be really conservative driving it (just the grip followed by backoff to get the pincer out of stall state while still maintaing a good tight grip takes a few seconds ).

Did you end up trying to drive your scorbot with a RoboClaw?

Cheers,
Michael

Hi Michael,

Haven't done anything with it yet, waiting to make deal on some RoboClaws!

I do have it moving with the stock controller, so that may have to do for a while.

Are you "talking" to the stock controller then?

Alan KM6VV

_________________
Visit:
http://groups.yahoo.com/group/SherlineCNC/
http://tech.groups.yahoo.com/group/HexapodRobotIK/

Yeah, I'm using the serial protocol to talk to the ER-III controller right now. Let me know if you have any questions on driving it via the serial port.

I put it into 'interrupt' mode most of the time (which basically just means it sends a single byte back whenever it does an internal state transition), but that doesn't let me ignore stall state on some motors (specifically the gripper), since I need the '0' response back to tell when it's finished a move and it'll never send that if any motor is in an error condition. I also can't adjust the PID parameters or the ramp up / ramp down curve to get smoother motion (though there is a command to set the speed per motor from 1 to (1)0 arbitrary units), so I could estimate how long a motion is going to take and keep changing the speed, but that seems like a bother). Because of all these limitations, I'd like to get a more flexible motor controller to handle this stuff, hence the interest in RoboClaw

My software doesn't do proper motion planning and I've never even measured the linkages to build a FK/IK model yet. Instead I've recorded delta encoder counts from home for a bunch of points of interest, plus graph edges (legal safe transitions from point to point that won't hit a fixed piece of wood, etc...), and it just does a graph traversal to find a safe route between any two points.

Even though I bought it used and beat up, it's super repeatable once I've got it homed right. However, my automatic homing is not very accurate (the limit switches are like 30 encoder counts wide; and there is only one limit switch for the combined roll/pitch of the end effector), so I get it close then move it to a fixed coarse registration point, jogging it to be correct for the fine registration, then I 'back out' the delta between the 'jogged position' and the 'known good fine registration position' from the home position. I always move it back to home after each use, so I only have to re-home it if I'm working on the table and I bump the arm while it's not energized or it hits something and a belt slips.

Cheers,
Michael

Sounds like you've done quite a bit with it Mitchael!

I'd be curious to hear more about your software. I've only ran the "SCOR1-4" programs, doing some simple moves (got sidetracked, as usual). I got a good buy. arm is in good shape.

What language are you writing in?

Have you considered upgrading the encoders? Might even be necessary to run the RoboClaws. The installed encoders are really low-res, as you've observed.

There is another thread on "multiple RoboClaws", I suspect that would be important! Have to wait to see the outcome there.

I assume you know about the Scorbot group?

http://tech.groups.yahoo.com/group/ScorbotUserGroup/

Might even be advantageous to carry on a conversation there (I like being able to save my emails).

Alan KM6VV

_________________
Visit:
http://groups.yahoo.com/group/SherlineCNC/
http://tech.groups.yahoo.com/group/HexapodRobotIK/

If you want to run multiple roboclaws and you have a problem running them bussed)eg all on one serial pin) then you can use 2 seperate serial pins instead. Currently we know someone that is having trouble with the older pre-RevC boards trying to use a single bussed pin. The Rev C boards should work fine with a single bussed pin but I am going to test them again to be sure.l

_________________
Tech Support
Basic Micro - Robotic Technology Evolved