Environmental Engineering Reference
In-Depth Information
Plan
Testing
Learning
Techniques
Planning
Technologies
Simulation
Software
Cooperation
Languages
Testbeds
Reasoning with
Partial Information
Agent
Technologies
Perceive
Act
Image
Processing
Robotic
Actuators
Data
Fusion
Communication
Signal
Processing
Fig. 7.2.
Cooperative autonomy technologies
When designing these agents, issues such as how the domain is divided be-
tween agents, how the agents negotiate among themselves, and the degree to
which they cooperate must be resolved.
Cooperation occurs whenever autonomous agents work together to achieve
common goals. Cooperation can occur in a variety of ways depending on the
domain and goals being pursued. The following sections will describe sev-
eral different patterns of cooperation and show them in relationship to the
Plan, Act, and Perceive cycle described above. These patterns are not mutu-
ally exclusive, and often the agents must simultaneously engage in multiple
cooperation patterns.
Cooperative Planning
Autonomous agents that work in groups as peers use cooperative planning.
They cooperate with one another to reach agreement on the actions each
member should perform to achieve their common goals. Figure
7.3
depicts a
group of autonomous agents that are engaged in cooperative planning. Each
layer of the Plan, Act, and Perceive cycles represents a separate autonomous
agent and the dots represent the points where these layers communicate dur-
ing cooperation. In cooperative planning, the communications are focused on
exchanging local views of the world, local and shared goals, actions that can
be performed, and shared priorities. During cooperation, each member must
reach an agreement on the set of actions that it will perform. In this pat-
tern, the agents exchange information during the planning activity and before
action is taken.
This form of cooperation is peer to peer. No single agent needs to be
in charge, and yet, each agent participates in a global negotiated solution.
