In this paper, we present a set of visual related routines. Our object
ive is to provide a mobile robot with the visual capabilities necessar
y to acquire information to execute a given command. We study a generi
c problem that we call the pickup and delivery problem, in which the r
obot is commanded to go to some location, pick up an object, and deliv
er it somewhere else. The concept of visual routines consists of havin
g a set of primitive routines that, combined in a specific way, result
in the exhibition of a complex behavior. Consistent with this idea, o
ur approach is to develop routines to cope with some visual problems t
hat arise when a mobile robot navigates in a given environment. Among
others, we deal with issues that arise during robot self-localization
and obstacle avoidance. A prime assumption that we make is that the en
vironment provides a fixed referential frame. The robot uses this info
rmation to adjust itself to the given environment. (C) 1998 Elsevier S
cience Ltd. All rights reserved.