Environmental Engineering Reference
In-Depth Information
in the organization, or from the ground, and execute them. An example
includes receiving and executing a control command sequence to move to
a new position within the cluster. This type of intelligence is similar to
that being flown on most spacecraft today.
The next higher spacecraft-level agent is I3, with local planning function-
alities onboard. “Local” means the spacecraft-level agent is capable of
generating and executing only plans related to its own tasks. An example
would be trajectory planning for orbital maneuvers.
Agent I2 adds a capability to interact with other spacecraft-level agents in
the organization. This usually requires the agent to have at least partial
knowledge of the full agent-based organization, i.e., of other spacecraft-
level agents. It must, therefore, continuously keep and update (or receive)
an internal representation of the agent-based organization. An example in-
cludes coordinating/negotiating with other spacecraft-level agents in case
of conflicting requirements.
The spacecraft-level agent I1 represents the most “intelligent” agent. The
primary difference between I1 and the other spacecraft-level agents out-
lined is that it is capable of monitoring all spacecraft-level agents in the
organization and planning for the organization as a whole. This requires
planning capabilities on the cluster level (a cluster being a subset of a con-
stellation), as well as a capability by which an agent has full knowledge of
all other spacecraft-level agents in the constellation. An example includes
calculation of a new cluster configuration and assigning new satellite po-
sitions within the cluster.
Selecting the level of intelligence of a multiagent organization is a complex
design process. The organization must be:
Adaptive, able to avoid bottlenecks, and able to reconfigure
Ecient in terms of time, resources, information exchange, and processing
Distributed in terms of intelligence, capabilities, and resources
A design selection process starts from an initial spacecraft-level intelli-
gence hierarchy. An example used for TechSat21 [ 133 ] is shown in Fig. 9.3 .
Here, high-level mission tasks were decomposed into lower-level tasks. The
spacecraft functions required to support these tasks are listed down the left
hand column with subfunctions grouped by category. Across the top are high
level spacecraft tasks with subtasks underneath. Tasks are then arranged in
a matrix form to provide a visualization of the agent capabilities associated
with each level of spacecraft intelligence as presented in Fig. 9.2 . The boxes
contain the ID's for function and subfunction categories.
9.5.2 Multiagent-Based Organizations for Satellites
Figure 9.4 shows a summary of options as a function of individual spacecraft-
level agent intelligence. Note that lower level functional agents are implied
in each of the architectures. As can be seen, the number and composition
Search WWH ::




Custom Search