Page 215 - Introduction to AI Robotics

198
6 Common Sensing Techniques for Reactive Robots
However, they wouldn’t necessarily be equivalent in performance or update
rate. As will be seen in this chapter, the sonar is liable to produce a noisy
percept in a second or two, while stereo vision may take minutes. Even dif-
ferent stereo vision algorithms may produce different results on the same
data stream. Therefore, the logical sensor contains a selector function which
specifies the conditions under which each alternative is useful and should be
selected.
Notice that a logical sensor can be implemented as a perceptual schema,
where the methods are the alternative means of generating the percept and
the coordinated control strategy contains the knowledge as to when a par-
ticular method is appropriate. Also note that each individual method can
be implemented as a perceptual schema, leading to the recursive, building-
block effect.
In reactive systems, the term logical sensor has degenerated somewhat
from its original usage and is essentially equivalent to a perceptual schema.
“Logical sensor” is often used to connote information hiding, where the par-
ticular sensor and processing algorithm is hidden in the “package.” This is
useful because a robot might use the same physical sensor in two different
ways. An avoid behavior might use a polar plot of sonar range data, while a
panic-stop behavior might use the minimum of all the incoming sonar data.
Since the perceptual schema use the raw sonar data differently, it is as if they
were different sensors.

6.2 Behavioral Sensor Fusion

SENSOR FUSION Sensor fusion is a broad term used for any process that combines information
from multiple sensors into a single percept. The motivation for sensor fusion
REDUNDANT stems from three basic combinations of sensors: redundant (or competing),
COMPLEMENTARY complementary,and coordinated. Although many researchers treat sensor fu-
COORDINATED sion as a means of constructing a global world model in a hierarchical or de-
liberative system, sensor fusion can be incorporated into behaviors through
sensor fission, action-oriented sensor fusion, and sensor fashion.
In some cases multiple sensors are used when a particular sensor is too
imprecise or noisy to give reliable data. Adding a second sensor can give
another “vote” for the percept. When a sensor leads the robot to believe that
FALSE POSITIVE a percept is present, but it is not, the error is called a false positive.The robot
has made a positive identification of percept, but it was false. Likewise, an
FALSE NEGATIVE error where the robot misses a percept is known as a false negative. Sensors