Biomedical Engineering Reference
In-Depth Information
Chapter 9
Applications of Neural Networks
in Micromechanics
A computer vision system permits one to provide feedback, which increases the
precision of the manufacturing process. It could be used in low-cost micromachine
tools and micromanipulators for microdevice production. A method of sequential
generations was proposed to create such microequipment [
1
], Chapter 8.
9.1 Neural-Network-Based Vision System for Microworkpiece
Manufacturing
According to this method, the microequipment of each new generation is smaller
than the equipment of the previous generations. This approach would allow us to
use low-cost components for each microequipment generation and to create micro-
factories capable of producing low-cost microdevices [
2
]. To preserve high preci-
sion of the microequipment, it is necessary to use adaptive algorithms of micropiece
production. Algorithms based on contact sensors were proved and showed good
results [
2
]. The neural-network-based vision system could provide much more
extensive possibilities to improve the manufacturing process.
As described in Chapter 8, a special project for microfactory creation based on
miniature micromachine tools was initiated in Japan [
3
]. The Mechanical Engineer-
ing Laboratory developed a desktop machining microfactory [
4
,
5
] consisting of
machine tools such as a lathe, a milling machine, a press machine, and assembly
machines such as a transfer arm and a two-fingered hand. This portable micro-
factory has external dimensions of 625 x 490 x 380 mm
3
.
The idea of microfactory creation is also supported in other countries. In
Switzerland, the details of precision motion control and microhandling principles
for future microfactories have been worked out [
6
]. One of the main problems in
such microfactories is that of automation on the basis of vision systems. There are
different approaches to constructing computer vision systems [
7
-
9
] for these
purposes.
