Aza’s Built to Code wrote:
Spit balling… Could optional vision based calibration assist web page help (e.g. javascript opencv, uses the phone’s Camera and CPU)?
not really, the problem with vision based calibration is what do you use as a
reference.
The author of WebControl tried to go this route with a camera in the router
collet referencing a grid. His biggest problem was getting an accurate grid. He
found that going to people who print banners and having them print the grid
resulted in a printout that was distorted significantly along the long axis
We have some ideas on how to proceed with this, but are not there yet.
see the discussions in
it turns out there has been some interesting work in the industry and academic
research on managing cable-driven robots and calibrating them. Ways that are
both faster and that better estimate accuracy than the current proccess (and may
even be able to figure out the Z offsets rather than us having to enter them)
Daivd Lang