Title of Paper: The Design and Implementation of Draper’s Earth
Phenomena Observing System (EPOS)
Principal Author: Mr. Mark Abramson
Abstract: Current Earth observation systems trade
coverage and revisit times against the number and measurement sensitivity of
satellites and on-board sensors. Traditionally, high launch and operation costs
drive the fielding of large multi-sensor platforms. The coverage and utility of
observations these large satellites provide are restricted by their selected
orbit, launch time, and bus reliability. In order to lower development and
operations costs, a recently proposed alternative is a suite of smaller size
satellites that perform single-purpose measurements. Within this concept,
however, high visitation rates of specific locations require enormous numbers
of platforms since the area of interest for observing short-lived events cannot
be predicted. Coverage and utility thus become restricted by the number of
satellites one can afford to place into the constellation, and visitation rate
is fixed by the constellation design. Earth observation is trending toward
adaptive systems of large numbers of multi-parameter or mission specific
satellites with the capability to provide continuous point coverage of
significant events as well as providing global monitoring to spot significant
events.
Draper Laboratory is currently researching methods to address how these
constellations or large numbers of platforms with heterogeneous capability and
availability can cooperate and be coordinated from the standpoint of
measurement utility, temporal concentration and resource allocation. This
methodology will enable many, independent systems to be used optimally for a
single task and to provide a scalable architecture as the number of available assets
grows.
We present initial results from algorithms developed for autonomous
reconfiguration of Earth observation satellites. This will reduce requirements
on the human operators of satellites, improve system resource utilization, and
provide the capability to dynamically respond to temporal terrestrial
phenomena. Our initial system model consists of small numbers of satellites, a
single point of interest on Earth (e.g., a volcano) with the objective to
maximize the total coverage of the target during a fixed time interval. An
approach using integer programming, network optimization and astrodynamics
jointly was used to calculate integrated observation and burn plans that
maximizes the total coverage of the target during the fixed time interval while
adhering to fuel constraints.