IEEE Robotics & Automation Magazine - September 2021 - 100

the satellite may suffer substantial damage and cease to function.
Additionally, many of these objects are large enough to
pose a significant threat to Earth. There are nonoperational
satellites in LEO, and they could enter Earth's atmosphere,
where they might or might not burn up. To remove such hazards,
these satellites need to be safely taken out of orbit.
Additionally, some asteroids and comets that orbit Earth
in the LEO range include valuable minerals that are no longer
frequently found. Thus, engineering missions intend to capture
and excavate them [24]. To lessen the threat posed by
space debris, collector satellites have been launched [25], but
their sensor range does not exceed a few kilometers. Moreover,
space debris detection using GPS satellites is not economical
or even generally possible. Debris collection requires
the spectrum analysis of particular electromagnetic signals,
which is beyond the capabilities of many satellites. In this
regard, astrobots, with their fibers that capture near-infrared
rays, have shown potential to contribute to space debris tracking
missions [26]. Objects rotate in LEO range at very high
velocities, causing friction in the thin atmosphere and
increasing their surface temperature. Thus, they radiate nearinfrared
waves that are detectable by near-infrared fibers [27].
One may note that a relatively large focal plane equipped
with these fibers can cover a vast area of LEO-range space
compared to what a single in-orbit satellite can do. This application
simply signifies that investment in astrobots and spectrograph-equipped
ground telescopes can potentially lead to
the robots' extensive use in many science and technology disciplines.
In summary, we argue that more astrobotics applications
may gradually emerge to contribute to scientific
discoveries and technological developments. Satellite observatories
have already contributed to the generation of spectroscopic
surveys [28]. However, because of the limited
capacities of these systems, their surveys are not comparable
to those of ground telescopes in terms of resolution. The
Training
Configurations
Training
Targets
Training
Results
Trainer
Test
Configurations
Test
Targets
Model
future miniaturization of astrobots may pave the way for generating
multifiber spectroscopic surveys in space.
Conclusions
Dark matter studies have been fascinating enough for astrophysicists
to seek a map of the observable universe. Cosmological
spectroscopy benefits from a plethora of optical fibers
whose configurations have to be updated from one observation
to another. Generating the desired map requires massive
numbers of optical fibers to be automatically coordinated.
Thus, the astrobotics field has emerged. Its design and operation
open a wide range of challenging theoretical and technological
problems. In particular, we explained how astrobots
constitute swarms of robots that exhibit unique issues stemming
from their specific requirements. In addition to the fairly
simple structure of a single astrobot, we addressed the
extremely challenging nature of coordinating astrobot
swarms so that collision-avoidance and completeness requirements
are fulfilled.
The target assignment problem was introduced, noting its
importance to and influence on the quality of the subsequent
coordination phase. The efficiency of optimal target assignment
was described in the case of massive spectroscopic surveys.
However, the key open question is how to localize
optimizations in small neighborhoods of astrobot swarms
during the course of assignment processes. Furthermore, with
smaller astrobots, more artifacts can be placed in focal planes,
generating a higher survey information throughput. In particular,
astrobots' stiffness and rigidity must be compensated for
on smaller scales. The development of microelectromechanical
system-based motors and bearings is also a must.
We also reviewed the two leading candidate methods to
safely and completely coordinate astrobots swarms: CNH control
and supervisory control. The CNH control strategy
targets the completeness of a whole swarm via the accumulation
of locally complete partitions.
We introduced two potential
options to optimally control coordination-dynamic
programming and
model predictive control-and we outlined
how machine learning may be
used to predict completeness of coordination.
The application of astrobots for
tracking space debris in LEO indicates
the prospect of further expansions of
the robots' applications in space and
astronomical studies.
Predictor
Predicted
Result
Coordinator
Verifier
Test Results
Figure 8. The flow of a potential coordination prediction process.
100 * IEEE ROBOTICS & AUTOMATION MAGAZINE * SEPTEMBER 2021
Prediction
Efficiency
Acknowledgments
This work was financially supported by
the Swiss National Science Foundation,
under grant 20FL21_185771, and by
the Sloan Astrophysical Research Consortium/Swiss
Federal Institute of
Technology in Lausanne, under agreement
SSP523.
IEEE Robotics & Automation Magazine - September 2021

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2021
