vision tests

Information and discussion regarding the ACE PackXpert software

vision tests

Postby TravisA » Mon Apr 21, 2008 4:01 pm

I started doing some tests to see the effects of one or more cameras, and whether or not robots are running.

I have uploaded the test results in excel format to the server at
\\srv-dc-1\corp\Projects\APM\SoftwareTesting\vision times

System: single PC, pentium4, 1GB RAM, ACE 1.5.4.5, standard vision(non remote)
Camera(s): Basler 1394, pylon driver
Image: White background, several black guitar picks on it
Data: Environment.TickCount stored to array, then saved to text file, 500 data points collected for each test.

Vision tool is run from a C# program in the ACE workspace.

The first tests use a blobfinder for the vision tool

Test #1
C# program with loop that executes the vision tool repeatedly, no delay
No robots running.
In this test the integer times were being converted to strings and appended to a string variable inside the loop. I didn't like this extra work happening inside the loop so I threw the results away.
Program was changed to only collect time values in the loop, then convert to string after all 500 pictures taken.

Test #2
C# program with loop that executes blobfinder repeatedly, no delay
No robots running.
-->all times range from 31 to 45ms

Test #3
C# program with loop that executes blobfinder repeatedly, no delay
2 robots running static part/target
-->times mostly in range of 31 to 47 ms, 2 results at 63ms

Test #4
C# program with loop that executes blobfinder repeatedly, no delay
4 robots running static part/target
-->times mostly in range of 31 to 47 ms, 9 results at 63ms

Test #5
C# program with loop that executes blobfinder repeatedly, no delay
4 robots running fast (high accels, no break at end of motion)
-->times mostly in range of 31 to 47 ms, 15 results at 63ms

Test #6
C# program with loop that executes blobfinder repeatedly, no delay
4 robots running fast (high accels, no break at end of motion)
ACE displaying vision window
-->times mostly in range of 31 to 47 ms, 31 results at 63ms

Test #7
C# program with loop that executes blobfinder repeatedly, no delay
4 robots running fast (high accels, no break at end of motion)
ACE displaying vision window and 3D Visualization window
-->times mostly in range of 31 to 47 ms, 37 results at 63ms, 1 at 78ms

Test #8
Changed the C# program. Instead of executing the tool repeatedly, it runs in a timed loop so it executes the tool every 250ms
No robots running
-->All times at exactly 250ms

Test #9
C# program executing blobfinder every 250ms
4 robots running fast
-->most times at 250ms, 1 result at 266ms

Test #10
C# program executing blobfinder every 100ms
no robots running
-->all times either 109ms or 110ms

Test #11
C# program executing blobfinder every 100ms
4 robots running fast
-->all times either 109ms or 110ms

Test #12
I added a second camera to the system, and wrote an identical C# program for it. Each program runs the blobfinder every 100ms
No robots running
-->most times either 109ms or 110ms, 1 result at 125ms

Test #13
2 C# program executing blobfinder every 100ms
No robots running
ACE displaying Vision window (ping-ponging between images)
-->most times either 109ms or 110ms, each program had 2 results at 125ms

Test #14
2 C# programs executing blobfinder every 100ms
4 robots running fast.
ACE displaying vision window
-->most times either 109ms or 110ms, 1 result at 125ms

Test #15
Changed the vision tool. I now created a Locator tool and model. The C# program is executing the locator repeatedly, no delays
No robots running.
ACE displaying vision window
-->significant variation in time for results. most results are around 188ms, but typical range is 172ms to 203ms, with many results as low as 156ms or as high as 219ms. one result was at 359ms

Test #16
C# program executing locator repeatedly.
4 robots running fast
ACE displaying vision window
-->most results ranging from 172 to 203ms, with several at 156 and 219ms

Test #17
Created a second C# program, calling a second Locator (on a second camera)
2 C# programs executing locator repeatedly
no robots running
ACE displaying vision window
-->results are widely scattered, most between 200ms and 300ms. some results as low as 156ms or as high as 469ms

Test #18
2 C# programs executing locator repeatedly
4 robots running fast
ACE displaying vision window
-->results are widely scattered. most between 200ms and 300ms, some as low as 156ms or as high as 469ms.

To me it is significant that tests #17 and #18 had almost identical results. The execution time of a locator tool is already dispersed and adding part/target allocation didn't cause an increase in times.
TravisA
 

Re: vision tests

Postby TravisA » Mon Apr 21, 2008 4:08 pm

Test #17 execution time graph
tst17_graph.jpg
tst17_graph.jpg (98.04 KiB) Viewed 10745 times



Test #18 execution time graph
tst18_graph.jpg
tst18_graph.jpg (90.78 KiB) Viewed 10746 times
TravisA
 


Return to ACE PackXpert

Who is online

Users browsing this forum: No registered users and 1 guest