Page 132 - Introduction to AI Robotics

115
4.3 Subsumption Architecture
more goal-directed actions such as mapping. Each of the layers can be
viewed as an abstract behavior for a particular task.
LAYERS CAN SUBSUME 2. Modules in a higher layer can override, or subsume, the output from be-
LOWER LAYERS haviors in the next lower layer. The behavioral layers operate concur-
rently and independently, so there needs to be a mechanism to handle
potential conflicts. The solution in subsumption is a type of winner-take-
all, where the winner is always the higher layer.

NO INTERNAL STATE 3. The use of internal state is avoided. Internal state in this case means any
type of local, persistent representation which represents the state of the
world, or a model. Because the robot is a situated agent, most of its in-
formation should come directly from the world. If the robot depends on
an internal representation, what it believes may begin to dangerously di-
verge from reality. Some internal state is needed for releasing behaviors
like being scared or hungry, but good behavioral designs minimize this.

4. A task is accomplished by activating the appropriate layer, which then
activates the lower layers below it, and so on. However, in practice, sub-
TASKABLE sumption style systems are not easily taskable, that is, they can’t be ordered
to do another task without being reprogrammed.

4.3.1 Example
These aspects are best illustrated by an example, extensively modified from
Brooks’ original paper 27 in order to be consistent with schema theory termi-
nology and to facilitate comparison with a potential fields methodology. A
robot capable of moving forward while not colliding with anything could be
LEVEL 0: AVOID represented with a single layer, Level 0. In this example, the robot has mul-
tiple sonars (or other range sensors), each pointing in a different direction,
and two actuators, one for driving forward and one for turning.
Following Fig. 4.6, the SONAR module reads the sonar ranges, does any
POLAR PLOT filtering of noise, and produces a polar plot. A polar plot represents the range
readings in polar coordinates, (r; ), surrounding the robot. As shown in

Fig. 4.7, the polar plot can be “unwound.”
If the range reading for the sonar facing dead ahead is below a certain
threshold, the COLLIDE module declares a collision and sends the halt signal
to the FORWARD drive actuator. If the robot was moving forward, it now
stops. Meanwhile, the FEELFORCE module is receiving the same polar plot.
It treats each sonar reading as a repulsive force, which can be represented