Smart
Cameras
Keep Their Brains on One Chip Continued
By Toshi Hori
Hardware
vs. Software
 In
general, a machine vision system is considered flexible if
a ser can change its function without upgrading the hardware
to adapt to new requirements, such as when a user changes
production lines, products under inspection or processes.
A software-oriented system is slower than a dedicated hardware
system ; however, a hardware system is more likely to be obsolete
when the application changes. Applications based on the zero
instruction-set processor offer the best features of both
hardware and software; speed, relatively low cost and flexibility
for various applications.
The
main objective of system and component development using the
new chip is to create a machine vision solution that does
not require programming. A "smart camera" system
would learn various tasks when a operator show it object samples
and teaches it which are good and bad parts.
This
arrangement allows camera manufacturers to make products that
can perform tasks only with hardware and flexible firmware.
A removable hard disk can collect image data and results for
furthering the system learning process. By using parallel
processing, recognition speeds can reach down to 850 ns -
without additional computer boards or software.
As
speed increases, so does the number of analyses that the system
can perform on each part per minute. These options are attractive
in today's industrial marketplace, where quality assurances
is on everyone's mind.
|
