August 1, 2019 at 9:11 am #108033
Has anyone tried using the lowrider mechanism as an XY digitizer? For example, if I have some plans that I want to convert to digital, I could move the low rider head with a pointer attached to find the relative location of points on my paper plan. Is there an easier way to do this? I want to use my lowrider to cut out boat pieces which I have paper plans for. Any thoughts or advice?August 1, 2019 at 9:16 am #108034
Would it be feasible to put encoders in line with the steppers? That is to use the timing belts to drive the encoders. Thoughts?August 1, 2019 at 10:51 am #108055
Interesting. It’s feasible but I think you will have to do a lot of work to get the encoder movements into a usable form.
You might be able to create a large format scanner using existing pieces, namely octopi with a plugin that takes pictures periodically. They use it for 3d print timelapses but you could probably get it to take a picture as it jogs in X and Y. If you focus a pi camera at a close distance you can get a high res picture of a small area, step over and repeat.
Stitching the images together shouldn’t be hard in principle, but practically I don’t know. The output could be an insane gigapixel image.August 1, 2019 at 11:06 am #108057
If you want to trace, then yeah, you need encoders. Quadrature encoders. Since the Low rider uses powered motors to keep the two sides aligned, it would be easy to get them misaligned if they weren’t powered on. So you would be fighting it quite a bit. You could put an encoder on each side, and take a smart average, based on the other encoder value. If you have infinite budget, or at least a lot of free time, you could have one side free wheeling, with an encoder, and drive the other to match it’s position. The trick is, any noise would end up fighting the user, so you’d need to spend a lot of time on the software smoothing it out, but making it responsive enough to follow the user without binding. It would be pretty cool though.
The camera thing would be neat, but you’d either need to take a ton of photos, or have a good camera calibration. Camera lenses that can see with a 90 degree FOV have a ton of radial distortion, so pixels at the edges aren’t square. Like a fish eye lens. You can correct for this, but you have to incorporate some opencv, and you’ll need to make measurements using the camera you have, because no two cameras will be the same. Even with a 90degree FOV, at 4″ above a 4’x8′ table, you’d get 72 photos. With a 60 degree FOV, at 4″ you’d need 288 photos. Those numbers might even be low, because the edges of the images will have fewer pixels per in. That would be something though. If you added some overlap in the images, you could even get 3D features (3D from motion or 3D from video). The hardest part of a 3D from video calculation is determining the position of the camera. If you had that, then you might be able to get really good results without a supercomputer.August 1, 2019 at 1:23 pm #108070
All good points.
Your talking through the encoder problem made me think of another approach: trace with an analog joystick! https://www.v1engineering.com/forum/topic/joystick-managed-by-marlin/ All the injected commands show up on the console, so if you record those, you can reconstruct the trajectory afterward.
If you’re feeling really brave you could even replay the commands to perform a cut. To me seems scary but I guess it’s still safer and more accurate than tracing a line with a handheld tool.
August 1, 2019 at 6:04 pm #108109
- This reply was modified 3 weeks, 1 day ago by Jamie.
As I think more about this, I’d I’m going to set up encoders, I might add week make a new mechanism with only 2 degrees of freedom. Then it would be more straight forward to convert the encoder counts to coordinates. Thanks for your good thoughts and ideas!August 1, 2019 at 6:23 pm #108112
You might start with thw MPCNC. The far side will track better than the low rider without power, even at larger lengths.August 1, 2019 at 7:09 pm #108115
True… I needed an excuse to build one anyway… 😉 Now that I have a sliver of building experience, if I build a Mpcnc, what controller would you recommend? Should I stick with a mini-Rambo or should I try another one?August 1, 2019 at 7:37 pm #108120
For the digitizer? Idk… There is very little you could use on the mrambo. For CNC, it’s a beast and the only reason to do the full rambo is independent drivers for dual endstops.August 1, 2019 at 7:50 pm #108127
I would like to build a MPCNC to use as a router and I see how it could also be more easily be modified to be a digitizer ad well. My question regarding the controller applies to the router application. Sorry for my scattered thoughts…August 1, 2019 at 8:28 pm #108130
The mini rambo is a great controller. The ultimachine ones Ryan sells are rock solid.August 3, 2019 at 5:32 pm #108289
I’m thinking that I can build a large 48″ × 96″) 2D digitizer by mimicking an arm (e.g., shoulder and elbow). With two encoders if I know the angles if the 2 joints I can get the coordinates at the end if the arm. I may do this and mount it on the far end of my low rider at center X. If both arm segments are 50″ between the joints, I can reach the whole table. If I do this, I’ll post pictures… I really only have one application that i would need it for so it’s not a high priority…August 3, 2019 at 6:59 pm #108299
FWIW, if it’s just for one project, you could attach digital angle guages and then just type the numbers in a spreadsheet. I’m all for making glorious contraptions out of inexpensive electronics, but I think it has to be considered.
August 3, 2019 at 7:03 pm #108301
- This reply was modified 2 weeks, 6 days ago by Jeffeb3.
You must be logged in to reply to this topic.