Articles, Blog

AI Planning for Space Supplement

December 3, 2019

Space missions and related areas have been
a fruitful source of applications for AI planning over a long period of time. One
of the earliest AI planners applied to spacecraft missions was Steve Vere’s
Deviser planner created at NASA’s Jet Propulsion Lab in the early 1980’s.
Optimum-AIV is a deployed planner used for Ariane rocket assembly, integration and
test and is based on the O-Plan design. Applications have also included ground
segment planning and control. Telecommand of meteorological satellites. Experiments
in technology proving and in the control of autonomous spacecraft. There are a wide
range of planners built at the European Space Agency and in NASA, and we’ll
cover some of those in this feature. And the Remote Agent Experiment within Deep
Space One. Devisor created at NASA’s Jet Propulsion Laboratory by Steve Vere was
the first NASA AI planner. It was based on the Nonlin design, but added time window handling.
It could produce Voyager mission plans. As part of our own work with the
British National Space Centre on the T-Sat technology proving satellite design. The
T-Sched scheduler was able to generate plans on the ground that were suitable for
controlling the spacecraft. A 24-hour plan was created, uploaded and executed live
onboard the UoSAT-II spacecraft. In early work at the European Space Agency, an
AI planner for ERS-1 operations planning was created, and later modified to become the
Optimum-AIV planner for assembly integration and test of the payload bay
of Ariane rockets. There are quite a range of projects at the European Space Agency
that use AI planning methods, and a few of them are listed here. You can find more
details on the European Space Agency’s spacecraft operations websites. ESA’s
project for on-board autonomy, PROBA, has had a series of spacecraft which have
demonstrated potential and feasibility of small satellites for advanced scientific
and Earth observation missions, and some have been controlled by automated ground
systems. NASA have also had a wide range of spacecraft with AI planning embedded in
them over the years. And some examples are listed here, such as ASPEN and
CASPER. More details of these projects are available at the NASA JPL and
NASA Ames websites. The NASA Mars Exloration Laboratory – Curiosity –
has been a target for a range of AI methods during its development stages and now in its
operation. This has not only included the rover itself, but also the communications
with it, spacecraft to support it, the operation centers which are used to
control and communicate with the robot, and even some of the preparatory work which
has gone into the design and simulation of the rovers and other equipment that’s
relevant to these missions. Take a moment to think about the Curiosity mission.
Imagine the opportunities for AI methods… from its design and inception as a mission
right through to the launching of the spacecraft, the landing of the spacecraft
on Mars, and then its operations.

You Might Also Like

No Comments

Leave a Reply