I’ve been working the past few years on a Mach 4 and Centroid compatible visual system for Avid. Works like a camera probe with a calibrated camera that closely maps what it sees to coordinates on your work surface. From there you can store useful positions, capture part geometry, then export DXF to CAD, or use built-in CAM to cut.
Here’s a video showing how to navigate by clicking in the camera view, find circle centers, cut what you see:
I’ve been mainly dealing with factory automation eg inspection processes for PCB manufacturing, but I’d like to hear what projects where it would help you.
To get people thinking about possibilities, I would like to have this to replace my probe and touceplate:
Setting X-Y center of features (squares, circles, between edges, etc.)
Corner finding
Bonus, find Z height
Setting the location to drill or cut a mortise on an existing part that I don’t have CAD for, or the workpiece is large and unusually shaped (like dropping a butterfly on a large slab crack)
Setting the X,Y, Z offset from the optically probed point to one of my lasers or my spindle (i.e. a library of user created offsets for different cutting tools that can be applied with a push of a button).
Mapping and converting to vector file different features or objects. Anyone working with live edge slabs can appreciate this.
Capturing the position of an odd shapped workpiece and aligning the CAD to it
Would be cool to be able to mark the bounds of an object. For example, I layout a mortise on a chair leg in pencil, then I mark the corners or line edges with the camera to define the boundary and then I tell it what bit I’m using and how deep to cut and it cuts the mortise for me.
Basically anything a physical probe can do, but you can also align it with crosshairs on the camera image so you can also you unprobable features like a pencil mark. Whats cool about this is you don’t have to install or adjust a probe, and then probe, and then remove the probe again, etc. you can just turn it on and do everything from the computer.
You should also be able to use it for squaring the gantry. It should be able to look at a known picture/item, and move around looking at the picture from different positions to know if the machine is square, and how much it needs to be adjusted.
Speaking of looking at things from different positions you should be able to generate models using photogrammetry. Though this will be hampered by no forth axis on the camera.
I used a camera on Mach 3 as a corner finder, and as a way to tell if my board was aligned on the axis. On Avid Ex I don’t really care about the alignment because I can use the WCS angle. Calculating that would be useful.
I see value in being able to set the Centroid coordinate system rotation (CSR) using the camera to identify the locations. Could be useful for avoiding flaws in wood or other times ultra precision isn’t needed to establish a good CSR.
You can both position and rotate GCode that you bring with the “GCode Operation”. You’ll see the bounding box of the gcode in the camera view and can move and rotate it. This will wrap your code with add a G52 that includes the camera offset from the tool and a g68/69 that rotates your gcode as you did in the camera view.
This chat has mostly focused on probing, detecting, alignment. I wanted to take a sec and show the cutting side of things since I introduced visual CAM operations. If you’re starting with GCode or just an SVG, you can quick go from vectors to cuts. The twist is you’ll see where the cut aligns, say you’re engraving, making inlays or just want to actually see what’s gonna happen. You can also aircut first and send output gcode to your machine to use graphing operations there. In short, I’m trying to close in on a very tight understanding of what you’re going to get before you put tool to material.
Here’s an example starting from an SVG file and using Slot operations with a v-bit to engrave and a cutter to remove, note the tabs too.
