About the hardware. I would print all the parts and assemble a working robot, then go back and make then with metal. If yu are going to find problems it is best to do this with a material that costs only 2 cents per gram.
If you don't want to paint yourself into a corner with software, then follow the conventions that "everyone" uses. The convention is to first create what is called simply enough called a "robot controller". This uses a specific kind of interface called "Joint trajectory Action" and when you look inside it is a series of points where each point has 1) a position for each joint (of the joint in radians or in meters depending if it is a linear or rotary actuator) 2) an optional velocity for each joint 3) an optional acceleration for each joint 4) a time (in seconds from some "start time" when the goal needs to be met All of the above are 64-bit floating point values The robot controller sends back the actual points that it was able to make. These points can come ahead of time and might get buffered and then the robot controller can do some lookahead planning or they might com one point just in time to be executed. Note that the robot controller does not worry about coordinated movement. If there are 6 joints, each is controlled independently. Coordinated control is done at the next higher level. This next higher level needs to understand forward and reverse kinematics and and motion planning that avoids collisions If you keep to this loose standard then you will be able to swap high-level controllers The higher level controller might be LCNC if your robot is going simple tasks like tool changing or loading stock onto a CNC machine. But if you robot is fetching beer from an unorganized refrigerator, LCNC would be a very poor choice for a controller. I think most robots use "MoveIt." you can see how it works here: https://moveit.ros.org/ On Sun, Feb 5, 2023 at 6:24 PM John Dammeyer <jo...@autoartisans.com> wrote: > Actually yes, the plan is to create a series of patterns and cast aluminum > blanks and then CNC machine them. > > Because the mechanical drawings are there too, and I've not been able to > use FreeCAD 0.19 to convert to STP file format, it's easier to recreate > the Alibre Part files from the drawings. > > For now what I've started is generating the G-Code for my 3D printer so I > can get a feel for pattern sizes. > > But it's the software in particular that is the interesting part. The > rest is really just mechanical assembly. > > Like most Arduino programs it's very linear and gets away with that > because of the processor speed and memory. But, as one comment to his > video stated, it's very slow and the suggestion was a change to Servos. > And yes, that would probably increase the speed. However, the way the > steps are generated with physical waits between steps, where nothing else > is done, is a major limiting factor too. > > I think a simple 10 MHz Spi bus shift register with the image of 8 step > pulses would make a major difference in the amount of code space used and > time wasted. But then a special board needs to be made. With the extra > I/O on the Teensy freed up it's possible to implement CAN bus and then run > other slower I/O devices with CANopen instead of a second USB port. > > So I'm back to wondering about the user interface and the teaching > capabilities and to use LinuxCNC instead of hard coded Teensy Arduino style > programming. > > John > > > > -----Original Message----- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: February-05-23 5:20 PM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Robot Arm control via LCNC, Pi4 and MESA. > > > > I have been following this for a long time. He makes the STL files > public, > > do what you like with then but STL files are really only good for 3D > > printing. He makes money by selling metal CNC'd version of the same > > parts. So you can build a plastic version of the arm for the cost of > the > > plastic (and motors and bearings) > > > > If you want to make your own CNC'd metal version, you could reverse > > engineer the STL files. The only parts that need to be redrawn are the > > bearing pockets. It is easy to reverse-engineer STL files. I am > doing > > this right now for an unrelated project in another open window with > > Fusion360. > > > > What software to use to control the arm depends 100% on what you want the > > arm to do. With robot arms, there are several classes of things to do. > > All of the interesting things can not be pre-scripted but use sensors to > > "see" the world that the robot interacts with. > > > > On Sun, Feb 5, 2023 at 4:08 PM John Dammeyer <jo...@autoartisans.com> > wrote: > > > > > Ran into this a few days ago. > > > https://www.anninrobotics.com/ > > > > > > > > > It's fully open source with the usual disclaimer that one can build it > for > > > fun or education but can't make them to resell. > > > Yeah right� > > > https://www.aliexpress.com/item/1005003688277794.html > > > > > > The AR3 version used a Teensy 3.5 with the trajectory calculations > done on > > > the PC and strings of co-ordinates sent out to the Teensy for the > step/dir > > > and encoder feedback. > > > > > > The AR4 upgraded to the Teensy 4.1 and moved all the trajectory > planning > > > into the module itself. > > > > > > Although I've registered on his website to get into the forum so far no > > > word. The videos also seem to be about 11 months old. Maybe the > > > Aliexpress version isn't pirated but just production and he's > abandoned the > > > support for open source. > > > > > > Anyway, there have been discussions that LCNC does this sort of thing > > > quite well. I'm wondering if it's worth the effort to change over. > The PC > > > program for teaching and running programs is pretty impressive. At > least > > > to me. > > > > > > Comments? > > > Thanks > > > John > > > > > > _______________________________________________ > > > Emc-users mailing list > > > Emc-users@lists.sourceforge.net > > > https://lists.sourceforge.net/lists/listinfo/emc-users > > > > > > > > > -- > > > > Chris Albertson > > Redondo Beach, California > > > > _______________________________________________ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
