Basic Laboratory Procedure
Unlike many areas of computer science, computer vision is a laboratory
science. Computer vision researchers frequently assemble and
manipulate experimental hardware, take physical measurements, and
collect and analyze experimental data.
Because it is hard to hurt yourself with a computer, computer
scientists do not usually have to pay much attention to safety. One
major exception, however, concerns robots and robot arms. Minor
failures in control hardware can cause normally reliable commercial
robots (e.g. Unimation PUMA arms) to move suddenly in a random
direction, even if no one has commanded any motion. This happens more
frequently with old robots and robots under development. Many robots
can deliver very large forces, enough to cause serious injury or even
death. At full speed, robot arms may move very fast, faster
than you can react.
Do not ever go into a robot's workspace when it is powered on.
Many computer science students have little experience taking scientific notes.
This can cause problems when they embark on projects with a substantial
experimental component, for which well-organized notes are essential.
Useful advice from a field with much more experience in this area
can be found in
- Kanare, Howard M. (1985) Writing the Laboratory Notebook, American
Chemical Society, Washington D.C. (ISBN 0-8412-0906-5 hardbound,
When the experiment is finished, it will be necessary to analyze your
experimental data. Some basic references on scientific data analysis are:
- Bevington, Philip R. and D. Keith Robinson (1992) Data Reduction
and Error Analysis for the Physical Science (second edition), McGraw-Hill,
New York (ISBN 0-07-911243-9).
- Taylor, John R. (1997) An Introduction to Error Analysis: The Study of
Uncertainties in Physical Measurements, second edition
University Science Books (ISBN 0-935702-75-X paperback).