Stuart - you're not crazy - at this point it doesn't look like genserkins accepts the theta parameter, thus it's most likely hardwired into the transforms. I lost braincells working on matricies over the past few months and Alex's work on genserkins was a godsend. I now have a Puma 500 up and running quite happily under genserkins - of which I'm forcing out only 3 joints of kins for our application. I would love,love,love however to reposition the theta parameter for Joint 0 in order to re-orient the xy plane to not emulate a vertical machine quill. That's not available....yet. Along with a couple others, my next step will be to dive back into that work and see if theta can be integrated as well. It is the least-used parameter, given a Puma-style-manipulator, so I'm not surprised it didn't get in for the first round. If you look at pumakins, it's STRICTLY Puma(R)-style and leaves very little room for other configurations - so generkins even at this level, is quite an accomplishment!
There is an amazing disconnect in the world of robotic arm manipulators - it was mentioned that Mach3 would have a tough time doing this - would you believe that Unimation (Puma), Yaskawa (Motoman), Kuka, Kawasaki, Fanuc FM (or whatever they changed their name to again this month) - and others - practically all arm manufacturers can't do what EMC2 does? In their world, a robot arm is strictly a parts manipulator. Teach-in and point-to-point programming - that's it. True, there's a Mastercam plugin for most platforms, and if you're really clever with VAL programming, you can write scripts that covnert xyz point tables into macros....but to directly take g-code and emulate a mill quill? I got a lot of "lost deer in headlights" looks, even from support reps at the booths in IMTS this year. I was specifically pushing them on this capability and their response categorically was "that sounds incredibly cool - we'd like to see it - but to operate your robot arm we have this pendant-thingie here. Or one guru back at headquarters that can write some XML for you." I think one of the big reasons it isn't pursued is due to the fact that understanding the work envelope versus the workpiece envelope is a challenge. The first thought of looking at a 6' tall arm is "wow...you can do a really big part with that thing" - we anthropomorphize the machine to our own arm/shoulder capabilities. On my Puma 500, (17" links, and is currently mounted to a wheeled 48x48" pallet!) I really only have a usable single-part work envelope of about 14"x14"x6" - why so small? The work envelope of the arm is spherical, and g-code operations are relative to planar surfaces - so I need to work within a planar surface that is tangent to the sphere of my robot. That space shrinks very very quickly. In a point-to-point program, (typical part manipulator) you can ignore the actual path as long as you don't crash. In g-code, the path is the whole purpose of the code. Thus if the arm is "almost" at a joint max extension, there may be a good chance you run out of length and fault. A fully extended arm can't extend any further. In a part-manipulator situation it is very easy to visualize where the conveyor and vice/chuck are - and if they are within the work envelope. If they are within, then there's little chance your application will fail. Since the kins has to resolve xyz into (any potential number of) solutions given current position, next position and mechanical limits, there is an incredibly high chance that eventually, it will fail. I think that's the reason that practically no one else out there is even attempting this (aside from research institutions). It always amazes me for what many perceive as a "Hobby" platform how absolutely cutting edge the stuff that is done here is! Ted. <Stuart Wrote:>Date: Wed, 17 Nov 2010 11:23:44 -0600 From: Stuart Stevenson <stus...@gmail.com> Subject: Re: [Emc-users] American Robot at MPM To: "Enhanced Machine Controller (EMC)" <emc-users@lists.sourceforge.net> Message-ID: <aanlktimjajak=wwaac=yazbujoqq19x8ig512+dk8...@mail.gmail.com> Content-Type: text/plain; charset=UTF-8 Gentlemen, In studying the D-H parameter I see the use of 4 parameters. I see genserkins http://www.linuxcnc.org/docview/html//man/man9/genserkins.9.htmlhas only three listed. Either I don't understand how 4 becomes three or something is not correct in the DH parameters description or genserkins. Any clarification would be greatly appreciated. Also, I saw a comment by SWP of a change made during the setup of the robot here. Clarification here would also be met with happiness. I only ask this as I am still uncertain of PC start success (I haven't used the hammer yet) and think I may be facing a start over from blank hard drive. I am off to find a hammer. :) thanks Stuart -- dos centavos ------------------------------ ------------------------------------------------------------------------------ Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
