Page 272 - Introduction to AI Robotics

255
6.11 End Notes
the morning. As I walked past the pool, one of my friends said, “Oh, Chris Brown
is over there!” I immediately turned to look for what I thought would be an older,
dignified author wearing a suit and tie. Instead, I got soaked by a tall, youngish man
impishly doing a cannonball into the pool. From that day on, I never assumed that
textbook authors were dull and dignified.

The tallest roboticist.
Tom Henderson at the University of Utah was one of the founders of the concept of
logical sensors. Henderson is also the tallest known roboticist, and played basketball
in college for Louisiana State University.

Stereo with a single camera.
Ray Jarvis at Monash University in Austraila came up with a clever way of gathering
rectified stereo images from a single camera. He used a prism to project two slightly
different veiwpoints onto the lens of a camera, creating an image which had a differ-
ent image on each side. The algorithm knew which pixels belong to each image, so
there was no problem with processing.

USAR and Picking Up the Trash.
While picking up trash seems mundane, most people would agree that finding and
rescuing survivors of an earthquake is not. The two tasks have much in common,
as illustrated by the work of Jake Sprouse, and behaviors and schemas developed
in one domain can be transferred to another. Sprouse was a member of the Colorado
School of Mines’ 1995 IJCAI competition team. The same vision program he wrote for
finding “red” was used to find “international orange.” Later, he extended the search
strategies to incorporate aspects of how insects forage for food. 108

Run over Barney.
The figures and materials on color histogramming used in this chapter were part
of research work conducted by Dale Hawkins in persistence of belief. His use of a
stuffed Barney doll started out from a class project: program a mobile robot to find a
Barney doll and run over it. This is actually a straightforward reactive project. The
Barney doll is a distinctive purple, making it easy for the vision system to find it. The
project allowed the programmers to use the flat earth assumption, so trigonometry
could be used to estimate the location to the doll based on the location in image coor-
dinates. Hawkins’ program was the clear winner, running completely over Barney. It
also gave him vision code that he could reuse for his thesis.

Reactive soccer.
Color regions are often used to simplify tracking balls (and other robots) in robot
soccer competitions such as RoboCup and MIROSOT. One amusing aspect is that
many of these behaviors are purely reflexive; if the robot sees the ball, it responds,
but if it loses the ball, it stops. Ann Brigante and Dale Hawkins programmed a No-