Hierarchies of behaviors can be constructed and coordinated with
great versatility.
NASA's Jet Propulsion Laboratory, Pasadena,
California
The Control Architecture for Multirobot Outpost (CAMPOUT) is a
distributed-control architecture for coordinating the activities of
multiple robots. In the CAMPOUT, multiple-agent activities and
sensor-based controls are derived as group compositions and involve
coordination of more basic controllers denoted, for present
purposes, as behaviors.
The CAMPOUT provides basic
mechanistic concepts for representation and execution of distributed
group activities. One considers a network of nodes that comprise
behaviors (self-contained controllers) augmented with hyper-links,
which are used to exchange information between the nodes to achieve
coordinated activities. Group behavior is guided by a scripted plan,
which encodes a conditional sequence of single-agent activities.
Thus, higher-level functionality is composed by coordination of more
basic behaviors under the downward task decomposition of a
multi-agent planner (see figure).
Robotics is a highly
multidisciplinary field that requires efficient integration of many
components (e.g., perception, mapping, localization, control, and
learning) that involve diverse representations, frameworks, and
paradigms (e.g., classical control theory, artificially intelligent
planners, estimation theory, data fusion, computer vision, utility
theory, decision theory, fuzzy logic, and multiple-objective
decision making). The CAMPOUT provides a conceptual infrastructure
for consolidating diverse techniques to enable the efficient use and
integration of these components for meaningful interaction and
operation.
The methodology of the CAMPOUT features a few
elementary architectural mechanisms for (a) behavior representation,
(b) behavior composition, (c) group coordination of teams, and (d)
interfaces among (a), (b), and (c). For the purposes of the CAMPOUT,
a behavior is defined and represented as a mapping from a percept
(defined here as a description of raw or processed sensory input) or
a sequence of percepts to an action or sequence of
actions.
The mapping assigns, to each possible action, a
degree of preference that ranges from 0 for most undesired to 1 for
most desired. This definition of a behavior is a general recipe that
does not dictate how the mapping is to be implemented. It does not
exclude implementation by use of a look-up table, a finite-state
machine, a neural network, an expert system, a control law, or any
other such means. Each behavior can be implemented using whichever
approach is appropriate.
Behavior composition is the
mechanism used for building higher-level behaviors by combining
lower-level ones. The activities of lower-level behaviors are
coordinated within the context of the task and objective of a
higher-level behavior. An explicit design goal of the CAMPOUT has
been to support not one but an arbitrary number of
behavior-coordination mechanisms (BCMs). The architecture can be
extended by incorporation of new BCMs.
Because different BCMs
often require different behavior representations, the CAMPOUT
involves utilization of a multivalued behavior representation that
is general enough for a large class of applications. BCMs can be
divided into two main classes: arbitration and command. The CAMPOUT
supports both classes.
In the CAMPOUT, the problem of coordinating a group of robots is
formulated as one of coordinating multiple distributed behaviors
across a network that includes more than one decision maker. In
behavior coordination, one is basically concerned with resolving or
managing conflicts between mutually exclusive alternatives and
between behavioral objectives. Because this is as true for
individual as for group decision-making, the difference between
individual and group decision-making is inessential, and both can be
studied in the same conceptual framework. Mechanisms that are
typically used for coordination of the behavior of one robot can
then be used for coordination of behaviors running on a network of
robots. Hence, for example, a control loop could use sensors on one
robot to drive a different robot.
This work was done by Hrand Aghazarian, Paolo Pirjanian, Paul
Schenker, and Terrance Huntsberger of Caltech for NASA’s Jet
Propulsion Laboratory. For further information, access the Technical
Support Package (TSP) free on-line at www.techbriefs.com/tsp
under the Information Sciences category.
In accordance with
Public Law 96-517, the contractor has elected to retain title to
this invention. Inquiries concerning rights for its commercial use
should be addressed to:
