Space Drones: An Opportunity to Include, Engage, Accelerate, and Advance

1Georgia Institute of Technology, Daniel Guggenheim School of Aerospace Engineering, Atlanta, GA, USA. 2Georgia Institute of Technology, School of Earth and Atmospheric Sciences, Atlanta, GA, USA. 3ETH Zurich, Institute for Dynamic Systems and Control, Zurich, Switzerland. 4MIT Department of Aeronautics and Astronautics, Cambridge, MA, USA. 5MIT Media Lab Space Exploration Initiative, Cambridge, MA, USA. 6Monterey Bay Aquarium Research Institute, Moss Landing, California, USA. *Primary author: cecarr@gatech.edu, +1-617-216-5012.


Accelerate the development and applica on of autonomy to space missions by
leveraging a broad range of crea vity, insight, exper se, and perseverance. 2. Meet the moral impera ve of inclusivity , ensuring that NASA's mission 5 , to "pioneer advances in aeronau cs, space explora on, science, and technology to transform our understanding of the universe, unlock new opportuni es, and inspire the world" is not only for the people but also by the people. 3. Engage individuals and give them agency to create an autonomous future on Earth and in space that represents the full breadth of our be er selves. 4. Broaden par cipa on in aerospace and space explora on through the integra on of educa on and research, including enhanced par cipa on in space missions.
Stakeholders in planetary science, and more broadly, robo c and human space explora on, should leverage and extend the autonomy revolu on to enhance mission science return, our future in space, and human health and wellbeing. Addressing these challenges requires us to engage a global workforce, and create an environment where the seeds of ideas can become inven ons that grow and propagate. We believe that Space Drones, small uncrewed vehicles guided by remote or autonomous means , provide one compelling pla orm to achieve these ends. The feasibility of driving and flying on other worlds has now been demonstrated in the lab 1 , and will soon be demonstrated on the surface of another world [1][2] . With the helicopter Ingenuity now on its way to Mars, and the Dragonfly mission in development, the me is ripe to further develop a space autonomy ecosystem . This ecosystem is not new but can be strengthened by building upon exis ng programs and, where appropriate, crea ng new ones.

The Road to Inclusivity
Inclusion is a moral impera ve and a well-documented path to performance 6 . This is especially true in the context of planetary science and astrobiology, where solu ons require synthesis of mul ple disciplines and perspec ves. Whereas there have been field-specific improvements in diversity, in the geosciences, there has been li le progress in ethnic and racial diversity in the last 40 years 7 . Furthermore, diversity by itself is inadequate and unsustainable; cultures must also change to engender inclusion and belonging 8 . One approach to educa onal inclusion is to change the problem framework whereas another is to change the norms: Duckietown , below, reframes autonomy in terms that conform less to tradi onal gender stereotypes; as an example, women doing robo cs and leading space missions (GRAIL, Psyche, and Dragonfly, led by Maria T. Zuber, Lindy Elkins-Tanton, and Elizabeth Turtle, respec vely) are changing the norms and serve as role models. These and other approaches can lead to cultural change. Recent efforts by NASA to add inclusion as a core value, support conference codes of conduct, and provide resources for grant wri ng, mission proposal development, career development, and new PIs are notable.

Educa onal Paths to Enhancing the Space Autonomy Ecosystem
NASA and other stakeholders operate many programs suppor ng advancement in science, technology, engineering, art, math, and design (STEAMD), in order to achieve goals including innova on in science and technology. These programs naturally overlap with the themes relevant to a space autonomy ecosystem . We describe a few examples.
Massively Open Online Courses (MOOCs) regularly exceed tens of thousands of enrollees for a single course. EdX has 14M learners and 50M registra ons. Although comple on rates are o en low (5%), shorter courses are completed at higher frequencies 9 . MOOCs enable inclusion by crea ng o en-free educa onal experiences that can lead to cer ficates, but also exclude those without rou ne high-speed Internet access.
Aerial Robo cs ( h ps://www.coursera.org/learn/robo cs-flight ) , a MOOC taught by Vijay Kumar (University of Pennsylvania) has 95k people enrolled as of the August 2020 start date. This and other courses use micro air vehicles to teach flight mechanics, control theory, trajectory planning, and integrate learning by doing.
Duckietown ( h ps://www.duckietown.org/ ) is a state-of-the-art open-source ecosystem 10 geared towards teaching autonomy using low-cost self-driving cars (duckiebots) and (duckie)drones, that func on in a rubber-duckie populated smart urban environment, focusing on vision as means of percep on. This playful approach to learning autonomy is geared towards mul -modal autonomous vehicle opera ons, and addresses challenges from the component to system levels, to open scien fic ques ons in embodied AI 11 . These challenges can be varied to teach entry-level autonomy 12 as well as pose research-grade problems 13 .
The NASA Community College Aerospace Scholars (NCAS) program, geared towards community college students, integrates a five-week online course with a four-day engineering workshop at a NASA center ( h ps://nasaostem.okstate.edu/ ). It incorporates team building, networking, as well as an engineering design challenge (e.g., a rover compe on), and has involved students from 141 Minority Serving Ins tu ons (MSIs) to date.

Compe ons as a Driver of Innova on in the Space Autonomy Ecosystem
Recent efforts to drive innova on through compe on include the XPrize and MIT Solve. NASA and other organiza ons also conduct a diverse set of compe ons oriented around different aerospace or explora on themes including disciplines, themed or mul -theme hackathons, and integra ve efforts such as vehicle challenges (underwater, aerial, surface, human rover).
These compe ons ( Figure 1 ) foster innova on directly and indirectly, by allowing individuals to get excited, build their personal and professional network, develop skills, envision career paths, meet mentors, experience the thrill of accomplishment and, cri cally, fail quickly , and o en with low stakes. These are examples and the list of compe ons above is not inclusive. Many compe ons are built around Earth-based themes and could be adapted to incorporate "Space Drone" themes either as an op on (e.g., Space Apps) or as an annual topic (e.g., Ver cal Flight Society). Many other relevant ac vi es exist including analog missions 14 .

Student and Public Par cipa on in Space Missions
Aerospace and Planetary Science degree programs o en incorporate space mission design in senior undergraduate or graduate courses. The Planetary Science Summer School (PSSS) at Caltech/JPL ( h ps://pscischool.jpl.nasa.gov/ ), geared towards doctoral students through junior faculty, provides a hypothesis-driven mission design experience intended to develop future space mission leaders 15 . Students and the public have par cipated in space missions directly by reques ng imaging of Mars via the HiWish program 16 as well as controlling a camera onboard the GRAIL spacecra orbi ng the moon via the MoonKam program 17 .
NASA engages students in space missions through full life cycle (conceive, design, implement, operate) opportuni es including stratospheric balloon flights (e.g., LSU's High Al tude Student Payload program), sounding rockets, and CubeSats. Student-developed payloads are represented in mul ple space missions. For example, the Regolith X-ray Imaging Spectrometer (REXIS) 18 on the OSIRIS-REx mission is student-run, with the goal to help teach the next genera on how to design and create spaceflight hardware.

Accelerate the development and applica on of autonomy to space missions
• Support both mission-capability driven development and blue-sky open-ended work.
• Evaluate augmenta on of exis ng NASA R&D programs to facilitate addi onal development of small vehicle autonomy including drones. • Leverage contribu ons from all stakeholders, including industry.
• Provide explicit opportuni es linked to the integra on of robo c space explora on and human explora on and habita on on the moon and Mars. • Provide reference designs to frame research in specific areas.
• Build Space Drone themes into exis ng compe ons where appropriate. Engage individuals and give them agency • Support many entry points into the space autonomy ecosystem, including across the en re educa onal spectrum. • Build an array of opportuni es, across different me scales, so that individuals can benefit from, and contribute to, the space autonomy ecosystem over a sustained period. Examples include a MOOC, hackathons, student project/research support via SpaceGrant or other mechanisms, graduate funding via FINESSE, and other mechanisms. • Provide a web of con nuity between individuals, teams, and opportuni es, so that people can navigate from one experience to another; common transi on pathways should be evaluated and supported through exis ng or new mechanisms.

Broaden par cipa on in space explora on
• Provide payload opportuni es as part of NASA and industry-related space ac vi es.
• Include "Space Drone" opportuni es as an explicit part of ongoing planning for combined human and robo c explora on on the Moon. • Include "Space Drone" opportuni es as part of human Mars explora on robo c precursor missions extending through human explora on and habita on on Mars.

Pipeline Enhancement
Massively Open Online Course. We envision a Space Drone themed MOOC that could help par cipants learn relevant skills, and carry out projects related to future capability needs assessment, technology evalua on, skill development, and teambuilding. A short course (~5 weeks) may help support higher comple on rates than a long course. Strategic skill enhancement and team building are likely more cri cal than comprehensive coverage, and the MOOC could help develop domain-specific skills as a feeder to exis ng or new compe on(s). Extending Exis ng Compe ons: Many op ons exist; here we men on two possibili es. NASA selects many topics for its annual Space Apps hackathon and could further incorporate space autonomy to facilitate idea on and team building. A Mars or Titan drone theme could be selected for a future Ver cal Flight Society (VFS) compe on. Compe ons could be aligned with Mars 2020 ac vi es as Perseverance is slated to land on Mars in February 2021.
Space Drones Compe on: We envision a new compe on that would include both domain-specific and integrated interdisciplinary challenges. Teams could be constructed on an open basis or have ins tu onal, interna onal or na onal team sub-compe ons (think Robot Olympics: the 2021 Olympics will have a strong robo cs component). Sub-compe ons could also require specific types of cross-group collabora on, such as a system built by mul ple teams across geographic or interna onal boundaries, as is common in space instruments and missions.
Taking the example of a Mars micro-rover or helicopter, a compe on could include separate sub-compe ons focused on structural op miza on, obstacle avoidance, path op miza on, energy storage or transfer, rotor design, packability and deployment, sampling (accessing special regions), planetary protec on (steriliza on, sample handoff), autonomous mission planning, sensor technologies (imaging, ice prospec ng, weather), mapping, facility monitoring, steep slope access, and other applica on-specific goals that would support robo c and human explora on on the surface of Mars. Metrics used to evaluate each sub-compe on could include mass, energy, power, volume, me, accuracy, and other relevant measures.
Aside from these domain-specific ac vi es, integra ve challenges could be u lized in a similar way as by the DARPA robo cs challenge (2012)(2013)(2014)(2015). NASA operates simulated planetary landscapes (e.g., JPL Mars Yard I & II, JSC Rock Yard), to facilitate research and tes ng of payloads. Related facili es are used in NASA compe ons such as the NASA Human Explora on Rover Challenge. JPL and other organiza ons also have thermal vacuum chambers that can be used to simulate Mars surface condi ons, where flying is extremely challenging due to the low atmospheric pressure (~7 mbar), similar to stratospheric condi ons on Earth.
Compe ons may u lize physical or virtual environments and modes of par cipa on. The ongoing pandemic has highlighted both limita ons and benefits of virtual events. One approach to integra ng physical hardware and virtual events is to send kits to par cipants or have compe on entries sent to a central tes ng loca on; code/simula ons can be run remotely. Simula ons u lizing a probability distribu on (e.g., Monte Carlo) to represent uncertain parameters (e.g., landscape, atmosphere), would help to provide more accurate measures of system performance.
Stratospheric Balloon Model for Mars Rotorcra . Stratospheric balloons are rou nely used to deploy science payloads and elevate the technology readiness level (TRL) in prepara on for space missions. Stratospheric tes ng of Mars rotorcra technologies is a logical next step a er ground-based thermal vacuum chamber tes ng due to Mars-like pressures, and in some cases, temperatures. Deployment and recovery strategies would need to be refined.

Mission Par cipa on
Drones on the Moon. At the MIT150 Symposium on Earth, Air, Ocean and Space: The Future of Explora on symposium, Jennifer Buz presented SatellitePlay, a concept for gamified robo c lunar explora on that envisioned anyone on Earth with a video game controller and internet access being able to drive micro-vehicles on the lunar surface. The video game market now exceeds $150B/year (2019). We envision a "sandbox" where micro-rovers can be u lized for educa onal and research purposes. The benefits of such a program should be evaluated for inclusion in Project Artemis as one approach to integra ng robo c and human explora on.
Human Mars Explora on. SpaceX is currently developing Starship, with a projected Mars landing capability of 100 tons (91,000 kg or 100 Perseverance rovers). There are many steps to successfully landing such a vehicle on Mars, yet SpaceX, BlueOrigin, and others are making significant progress towards delivering significant mass to the surface of the Moon, Mars, or other worlds. We envision robo c or human missions bringing micro-drones or improved versions of Ingenuity (see Mars Science Helicopter posi on paper).
A fleet of small vehicles could integrate STEAMD ac vi es across academia, industry, students, and the public, crea ng a Mars "sandbox" for autonomy. The fleet could map the nearby rock distribu on to cen meter-scale, ground-truth ice and other poten al resources, serve as a network of weather sensors, and validate technologies such as autonomous opera ons, 5G/high-speed mesh network communica on, wireless power delivery, and develop new capabili es. Access to (candidate) special regions (regions where life as we know it might be capable of reproducing) could leverage highly sterile micro vehicles for sampling and transport of samples to a clean yet accessible loca on, facilita ng acquisi on by a larger rover or human, "breaking the chain" and reducing the risk of forward contamina on. Such a pla orm would help advance space autonomy at the agent and fleet level.

IMPACTS
We seek to democra ze access to space, and through this process, unleash new space mission capabili es. Developing a strong pipeline, beginning as early as elementary school, is key to the health of the fields of autonomy and machine learning 19 . We recommend longitudinal tracking of impact on par cipants including in comparison to a control group, such as done by the FIRST Robo cs Compe on. This approach supplements enrollment and comple on data and has provided strong evidence of impact (FIRST female alumni are 5X "more likely to declare majors in engineering and computer science than their peers"). Other measures include research experience, doctoral program enrollment/comple on, fellowships, and mission par cipa on. Self-assessment of skills, confidence, professional network, and measures of belonging can be used in combina on to assess the impact of inclusion efforts 20 . Assessment should also include technology-centered metrics such as publica ons, grants, patents, licensing, cita ons, downloads of research products, and (code) forks.
A robust space autonomy ecosystem will also facilitate pathways between space and non-space applica ons, and benefit stakeholders including par cipa ng individuals, academia, government, and industry. Expanding opportuni es within this ecosystem, including across interna onal borders, can help to rebuild a pro-science reputa on of the U.S. and improve collabora on with our interna onal partners. By developing a robust space autonomy ecosystem now, we are inves ng in a future full of new space mission capabili es ( Figure 2 ). Advances may arise directly from early-stage programs and emerge from later ac vi es of par cipants. Through incremental and transforma onal change, we can achieve a future where autonomous vehicles drive, swim, or fly longer, further, more autonomously, and coopera vely. We will expand access to steep slopes, fresh impact craters, caves, and other challenging terrains of geological and astrobiological interest, as well as providing valuable mapping, monitoring, and surveying capabili es to support crewed ac vi es on celes al bodies such as the Moon and Mars. Poten al impacts described by other posi on papers include Mars surface relay networks, a Mars Science Helicopter, aerial surveying of Mars remnant magne sm, volcanic cave explora on as a priority for astrobiology, planetary protec on for human Mars missions, aerial pla orms for Venus, autonomy for Ocean World explora on, and opera ons concepts for asteroids and comets. Building an inclusive space autonomy ecosystem is a cri cal enabler of exploring the cosmos as well as seeking and suppor ng life beyond Earth.