Component segregation application

Component segregation application

Postby sherif1tce » Tue Aug 02, 2016 4:13 am

Dear Sir,

We are looking for component segregation based solution using adept sight software. We have about 650 models of component . Right now , we have mixed lot about 160k consisting of this 650 models . We tried 5 models as sample . We found that time took about 200 milliseconds for each component . We like to know that

1. Is it possible to teach all 650 models in the software ?
2. How much time will be taken if a trial image compares with 650 master image ?
3. Can we use adept sight software without connecting Robot ?
sherif1tce
 
Posts: 24
Joined: Thu Mar 13, 2014 3:01 am

Re: Component segregation application

Postby Guenter Grass » Tue Aug 16, 2016 5:06 am

Dear sherif1tce,

I have not heard of any customer doing this type of vision "inspection" using AdeptSight. The highest number of models I have used in one Locator is 9.

To answer your questions
1. Is it possible to teach all 650 models in the software ?

It is possible to teach this amount of models. The size of the workspace will be huge. If you want to test this make sure to place the bounding box close to the model and to "CROP" all models in the model editor. This reduces the size of the workspace file as well as the size of the model in RAM.
How big are the models? I doubt that 650 models will fit in RAM at the same time since the Vision Library in AdeptSight can only use address space in the first 4 GB of RAM (32 Bit).

2. How much time will be taken if a trial image compares with 650 master image ?

I don't know. The default timeout of the vision tools is 15000ms

3. Can we use adept sight software without connecting Robot ?

Yes.


I doubt that AdeptSight is performing well under these conditions. I know of customers that have more than 250 products but they always know which part they produce and their locators always contain no more than 3-4 models each.


Best Regards
Günter
Guenter Grass
 
Posts: 189
Joined: Thu Oct 16, 2008 4:34 am

Re: Component segregation application

Postby sherif1tce » Thu Aug 18, 2016 2:41 am

Dear Sir,

Thanks for your reply . Currently , I'm using adept ace with Ethernet communication camera and locater tool for this application . After creating about 50 locater models , execution time is around 300 ms . My requirement is to reduce the exection time since i need to create 650 models . Hence , kindly suggest right tool that will compare current model with all other master models and provide result such that similar to that X model .( X model is one of 650 master models ). I dont need position X,Y and angle result at all . Pls provide the necessary support asap .
sherif1tce
 
Posts: 24
Joined: Thu Mar 13, 2014 3:01 am

Re: Component segregation application

Postby Guenter Grass » Thu Aug 18, 2016 8:23 am

Dear Sir,

there is no other tool to find models in AdeptSight.

It might be possible to reduce the run time of the locator a little by changing some of the parameters but the first thing is to see if it can run at all with 650 models.
What time do you need?

Best Regards
Günter
Guenter Grass
 
Posts: 189
Joined: Thu Oct 16, 2008 4:34 am

Re: Component segregation application

Postby sherif1tce » Thu Aug 18, 2016 8:34 pm

Dear Sir,

I can wait even up to 10 seconds to find the current model out of 650 models . After 50 models , we found delay during saving the workspace. What may be the reason ? . Also suggest right solution for the requirement
sherif1tce
 
Posts: 24
Joined: Thu Mar 13, 2014 3:01 am

Re: Component segregation application

Postby Guenter Grass » Fri Aug 19, 2016 1:22 am

Dear Sir,

concerning your questions.

After 50 models , we found delay during saving the workspace. What may be the reason ?

While ACE is saving a workspace there is always a delay.

Also suggest right solution for the requirement

I don't understand this question.
You already use the locator which is the correct tool to search for models. You have trained 50 models and the Locator performs very well (300ms). As I already stated now is the time to add your other models.
If one Locator can't do the job you can create multiple Locators with for example 50 models each. The disadvantage of this is, if the models are similar more than one Locator might match one of its models to the part. In that case you could compare the match quality of the found part in each Locator to each other and only take the one with the highest match quality. But typically one Locator with all models is always better.

Günter
Guenter Grass
 
Posts: 189
Joined: Thu Oct 16, 2008 4:34 am


Return to AdeptSight

Who is online

Users browsing this forum: No registered users and 1 guest