# Product Design: Automated Optical Inspection

Joined Aug 20, 2020
5
Hi,

I'm looking to build an open-source Automated Optical Inspection(AOI) device. This means the source code, hardware, and documentation would be released(non-profit).

You aren't sure what an AOI is? Then take a look at this Hackaday Blog Post on AOI(this was my source of inspiration for the project).

Now, my target user would be hobbyists, technicians, or engineers working independently or for small firms. I would appreciate it if you(the user that is) could take the time to answer my questions. Hopefully, with your answers, I can narrow down the scope of my project. If I'm lucky I can convince my professors to approve this as a valid school project.

Onto the questions:
• What did you like or didn't like in the AOI you've worked with?
• What features would you like to see in an AOI?
• What would you expect from an open-source AOI(i.e: features, speed, responsiveness)?
• How important are the GUI and its responsiveness(in the context of an AOI)?
• Would you prefer a desktop app or a web app?

As you can see, I'm trying to get insight into the user experience and see how important the user-machine interface for this sort of device.

Cheers!

P.S: I made sure that I'm respecting the forum rules. If I'm posting in the wrong subforum, then please re-direct me to the appropriate one.

#### MisterBill2

Joined Jan 23, 2018
8,808
The TS is asking for engineering that woud cost at least $20,000 (USD), and probably a lot closer to$40,000.
Good luck!

Joined Aug 20, 2020
5
The TS is asking for engineering that woud cost at least $20,000 (USD), and probably a lot closer to$40,000.
Good luck!
Sorry, but I don't understand your post. Are you trying to say that the product development will probably cost in the 20K-40K range? Or the market research will be worth 20K-40K? I'm really confused.

Nonetheless, I appreciate that you took the time to read my post.

Cheers!

#### MisterBill2

Joined Jan 23, 2018
8,808
Yes, this is certainly not a trivial project. "This means the source code, hardware, and documentation would be released(non-profit)." Developing the system, including those items, is what would be expensive because it will take time. Automated inspection or testing is not trivial, and I am presuming that when the TS says "optical" that what is really meant is video.
I was the electrical/electronics part of a great design tea for quite a few years and we created a large number of automated test systems that were used on assembly lines in auto plants, and other factories. So I have experience with these systems and what is involved in their design and creation.

There have been "smart cameras" available for purchase for quite a few years, and while some of the have very powerful software I am not certain that they are what the TS is thinking of.

#### andrewmm

Joined Feb 25, 2011
1,498
Hi,

I'm looking to build an open-source Automated Optical Inspection(AOI) device. This means the source code, hardware, and documentation would be released(non-profit).

You aren't sure what an AOI is? Then take a look at this Hackaday Blog Post on AOI(this was my source of inspiration for the project).

Now, my target user would be hobbyists, technicians, or engineers working independently or for small firms. I would appreciate it if you(the user that is) could take the time to answer my questions. Hopefully, with your answers, I can narrow down the scope of my project. If I'm lucky I can convince my professors to approve this as a valid school project.

Onto the questions:
• What did you like or didn't like in the AOI you've worked with?
• What features would you like to see in an AOI?
• What would you expect from an open-source AOI(i.e: features, speed, responsiveness)?
• How important are the GUI and its responsiveness(in the context of an AOI)?
• Would you prefer a desktop app or a web app?

As you can see, I'm trying to get insight into the user experience and see how important the user-machine interface for this sort of device.

Cheers!

P.S: I made sure that I'm respecting the forum rules. If I'm posting in the wrong subforum, then please re-direct me to the appropriate one.

The system is by definition, defined by the requirement,

What that means is a system that might be great for inspecting bottles on a production line has a very different requirment than one used to inspect silicon wafers.

They are both very similar, but very different,

An opensource system, what are you aiming for ?
or are you aiming modules that offer different degrees of resolution / speed / complexity ?

#### MisterBill2

Joined Jan 23, 2018
8,808
By the way, the TS is asking for product and service that I still charge for. So no hope of a "freebee" fro this quarter.

Joined Aug 20, 2020
5
The system is by definition, defined by the requirement,
That's why I'm reading research papers, looking around the web and forums for pieces of information to help me narrow down my scope.

What that means is a system that might be great for inspecting bottles on a production line has a very different requirment than one used to inspect silicon wafers.

They are both very similar, but very different,
You're right! I simply assumed and forgot to mention it's an AOI device for PCB fault detection.

An opensource system, what are you aiming for ?
or are you aiming modules that offer different degrees of resolution / speed / complexity ?
I wanted to make it open source(or non-profit) like this I avoid all the extra paperwork and licensing that my Uni would enforce on me(also an act of generosity and what not).

My aim is to make this an official Uni project, avoid busting my head with licenses and increase the speed of the inspection process.

I planned to build a very simple hardware platform for inspection(camera, lighting and platform). I love hardware,but it's not my specialty. My strong suit is software and I'm looking to implement image processing/computer vision techniques into the project. I'm currently building a simplified demo for one of my professors. I've built a simple 3D printed platform, slapped an ugly webcam on it, built a desktop GUI and some OpenCV behind all that. So far so good. Hopefully, my prof will approve.

If my description of the product seems vague, then you would be right. That's why I'm looking around for info to help narrow down the scope of the project. I'm even having a hard time defining the requirements.

#### andrewmm

Joined Feb 25, 2011
1,498
PCB inspection,
naked or populated PCB's ?

Naked PCBs relatively easy,
flat lighting across scene ,
your comparing against a relatively well defined golden sample, and no shadows.

the chose is , do you compare against a Gerber of the board, or a reference sample of the board.
a reference sample, relatively easy image normalisation and subtraction of one against other,
You can use the fiducials on the PCB to align the new boards image against the reference.

If the reference is a gerber, a bit more difficult, outline of the tracks need to be compared to define are tracks correct,
how to identify tracks and gaps as to narrow / wide, is more difficult, but not to much.

A populated board, now thats much more difficult,
spotting solder bridges / missing solder as well as all the above,
components miss placed, wrong way round, wrong component,

all whilst having shadows, and hidden parts

all getting very complicated.