Sightech Machine Vision - Self Learning Eyebot

Application - Bio Medical Pipette Quality Inspection

Trainable machine vision is used in this example to check the degree of flashing on pipette tips. PC-Eyebot calculates a score which is used here to judge the severity of the defects. The pipettes are then judged as PASS / MARGINAL / or FAIL. Note positioning is loose and not restrictive. Flexible output choices are offered: Screen display, relays, or RS-232 serial. Three relays can output PASS / MARGINAL / and FAIL, while the RS-232 port can output the exact score to your PLC., The images below show (left) a gold-standard pipette and (right) one with some flashing.

Step 1: Train machine vision system with gold standard pipettes. The orange "hits" below show the massive learning activity. After several gold standard pipettes are trained, PC-Eyebot is ready to inspect for quality.

Step 2: Vision system is ready to inspect for quality. During the inspection process, a score is continuously produced which represents the degree of defects. Comparing this score to three user settable bands allows the pipettes to be binned into PASS / MARGINAL / FAIL. The following image shows passing pipettes - one is a trained gold-standard with a score=0.00, and the other is a OK pipette with a score=3.31. You can see that they both PASS.

The following pipette is marginal with a score=9.00 which bins to MARGINAL:

This pipette is judged as a failure with a score=14.52 which bins to FAIL.

You can see that the pipettes can be easily graded with the score that is produced by the vision system. The whole process consisted of two major steps: Step 1 - LEARN and Step 2 -Inspect. It could not be any more straightforward.

Home | About Sightech | Contact | Products | Applications | Support | Technology | Careers | Site Map | Resources