AI & Smoothieboard

Hello there smoothie fans !

Just, a simple question, o rmaybe, a complicated question.

Do you know or can you suggest or do you have an example or post or video about how AI can be applied to a cnc machine (obviously, using a smoothieboard) ?.

For example… I suppose, a simple scenario can be applied to collect vegetables located over random positions. But, I saw this… and I want to know if more examples like this one (Sanding a Guitar Using PushCorp Force Control, a Kawasaki Robot, and Solomon Solmotion Technology - YouTube) are available… or if you have seen something similar… or… if you did an example working with AI and smoothieboard…

I don’t know if it’s correct to ask something like this… but, I would like to know more examples about this, and which kind of deep learning software was used to achieve these possibilities…

Thanks in advance…

OpenPNP uses AI (image recognition) to detect if components are correctly placed on the PNP machine, and uses Smoothieboard to move the components.

It’s definitely possible to use AI with smoothieboard, but Smoothieboard and AI have different jobs.

The way it works typically is:

[ camera / sensors ] → [ AI that reads images/sensors and makes decision based on data ( “intelligence” ) ] → Serial data (Gcode) → Smoothieboard → [ Motors / movement ] → Action → Go to first step, after the action is done, data is collected and a new decision is made, it’s a loop.

Does that make sense ?

Why do you ask? Do you have a specific project we could maybe help with, or is this just curiosity ?


Just I want to know how can be possible to move a robotic arm using dynamic distances for example (AI that calculates the distance between objects that are seen from a camera) and send these distances to smoothieboard and move as result, or, grind and pulish stainless steel tanks, where the shapes of the tanks will not be the same: height and diameter change for each tank, and obviously, the purpose is to know which kind of software can help to “predict” (as near as possible) the path to grind and polish, and send to smoothieboard.

Just, some examples for grinding and polishing over stainless steel: KW150 Grinding Head Out Circular - YouTube

I have no idea what kind of software would do that. You would likely have to either find a company that sells aid software (I’m not sure how to search for them), or you’d have to code all of this yourself.
Smoothieboard listens/interprets Gcode, which is pretty universal “robot movement” language, so if you find existing software it’s likely to be able to output gcode/talk to smoothieboard.


Yeah, sure, It’s clear that if I don’t know the software, “the path” (the gcode program) can be done manually and send to smoothieboard using the 6 axis, but, that was the reason why I was asking about the use of AI with smoothieboard. It seems too complex to write from scratch the software that can achieve that, but, for the moment, is curiosity. Thanks a lot for your time Arthur, and, I don’t want to close this question without asking what happened with smoothieboard v2. I saw the updates written to the backers on kickstarter, but I don’t know where is the site (if it exists) where we can see the docs related with that board (similar to the page done for v1, and the guides for each type of machine), or maybe, where we can see how smoothieboard v2 is working with projects related with the community. Hope to hear news very soon about that. Thanks in advance…

v2 is very close to release, there should be another KS update soon. The docs will be very similar to the v1 ones, and the fork should happen soon, but we are concentrating on production now.


That sounds great. I want to see the advantages with the projects built by the community using v2. Maybe you have to focus with the site showing the projects built with v2. Hope to hear very soon about the new site with new specs and more axes to be used. Good luck !