Tips to build kind of safety device


Im gathering ideas how to implment safety device for the cnc mill.
I have semi automatic cnc, and i need to control somehow that starting
job is safe. Example: someone forgot the work gloves on the table, cnc will be started remotely
and if the blade grips it, there will be big problem. It actually happened not once but three times.

First I implemented custom image difference detection program without any computer vision algoritms,
it works quite well, i can control environment lighting but not the color of wood, lately I have lot of false detections,
and every time a person must go and manually start the machine.

So my next idea is implementing a true machine vision. The system that can detect the presence of the material,
find foreign objects but makes difference of them - gloves or tools are dangerous but sheet of paper or pencil are safe.
I have a time to analyze and process image, even 5 minutes is totally okey.

Long story short, any advice building this system using openmv?

Yup! With TensorFLow and EdgeImpulse this is easy now. Just train an image classifier CNN to look at the table and detect if there’s a foreign object.

You just need to collect a dataset of objects on the mill and then when its clean and the CNN will do the job of detecting if there’s stuff on it.

Thanks for quick answer. But now i have more “sophisticated” error when trying to connect with edge impulse: Error creating SSL contect () both with log in or api key

That’s not something I can easily help with. That’s an external issue with your computer not being able to connect over HTTPS to the edge impulse website. Can you login to their website via an internet browser?


Yes I can log in and use edge impulse website. Right now I built dataset using edge impulse web so it’s ok now. More intresting is computer vision itself.