Object Tracking and Identification
How can a computer recognize the presence and identity of objects in
its environment? Many diverse technologies are being experimented with—some requiring
special markings or transponders and some that do not. Bar codes and
RFID (radio frequency ID)are perhaps the oldest and most developed. Bar codes are the
familiar Universal Product Code zebra patterns of lines found on many items, such as cans.
These codes can flexibly be read in any orientation by bouncing a scanning laser beam off
them. Researchers are currently experimenting with two-dimensional markings that will
allow the placement of much more information in denser displays and methods of
unobtrusively embedding the information in other printed material on the products. RFID
requires the placement of a small electronic device on each object that can then be read at
a distance without the requirement of visual access. Examples are the security tags
placed on store merchandise or library books that activate alarms
when someone passes through a portal. Passive tags, which can be produced very inexpensively, use resonant
electronics to reply when activated by the inquiring signals. More expensive active tags
require battery power, can provide extensive information when queried, and can be dynamically
updated. Researchers are working to decrease the cost and increase
the range and amount of information contained on the tags. For example, a Micron product
used for more complex inventory schemes such as tracking railroad
cars and trucks can be read at distances of up to three hundred meters. They even have a system that combines
cellular phone and GPS technology and will eventually allow items such as cargo
containers to be tracked anywhere in the world. Related technology has been applied
to tracking prisoners on house arrest, children at amusement parks (to prevent them from
getting lost), taxi cabs (to insure that drivers don’t take too lengthy breaks), luxury
automobiles (to track them if they are stolen), and military assets (in order to deploy
them in battle). Other researchers are attempting to increase the intelligence of
video-image processing systems so they can identify objects in natural scenes without
special markers or transponders. MIT’s Physics and Media group is
studying the possibility of avoiding circuitry and video altogether by using natural physical features of objects
and persons for tracking and identification. For example, the Spin Resonance project is
seeking to determine whether the atomic spin patterns of molecules could be sensed and
used for identification purposes.