We live in an exciting moment in aerospace history. Funding for space research is booming after years of stagnation, and the academic community is poised to fundamentally define the future of aerospace for the next decade. Our goal in the MLDS lab is to revitalize the community by exploring the most rapidly evolving facets of astrodynamics and computer science research to help define the next generation of spaceflight. We strive to be a leader in this space, investigating how novel machine learning algorithms can produce cutting-edge astrodynamics research.
Today, well-known machine learning tools like ChatGPT, DALLE, and others, are making headlines for their transformative capabilities for society. While these tools are indeed impressive, one could argue that even more exciting applications of AI exist in the science and engineering communities at-large. Researchers today are designing machine learning models which entirely redefine our capabilities in nuclear fusion, protein folding, and autonomous vehicles. It is clear that the early adopters of these technologies are making waves in their respective fields, but despite this, machine learning remains largely underutilized by many academic disciplines… including aerospace.
I argue this is because aerospace is a notoriously risk averse community, and rightfully so. There is no room for error when flying satellite worth hundreds of millions of dollars or carrying human payloads. Consequently, the community has be relatively slow to adopt, or even explore, these cutting-edge advances in machine learning. The MLDS lab is working to change this. Specifically, our goal is to design trustworthy and physics-based machine learning models that can be flown on-board spacecraft. We believe physics-based models are the crucial link that can unite machine learning with aerospace, producing models that are rooted in first-principles without sacrificing performance. At present, we are among one of the only research groups in the country exploring this intersection of work, and we’d love to have you join our team.
Leveraging top-tier facilities like the IDEA Factory and the Iribe Center, institutes like the Center for Robotics and UMIACS, and state-of-the-art high-performance computing capabilities, our students are equipped with the exact resources needed to excel at this research. Moreover, the graduates of the University of Maryland are exposed to a wide array of space-centric curricula instructed by faculty members and lecturers who actively work on flagship NASA missions, both on science and flight dynamics teams. Moreover, UMD has unique proximity to some of the largest aerospace research centers in the country, providing students a wealth of internship opportunities to position them well for successful aerospace careers.
My goal as an advisor will be to help you grow into a prominent researcher operating at this intersection of physics-based machine learning models and space systems research. I will work to give you the resources and mentorship you need to succeed here at UMD, offering consistent feedback on research projects, papers, and presentations. If you want to learn more about my advising philosophy and expectations, feel free to check out the FAQ below. I look forward to collaborating with you soon.
Current UMD Students
If you are a current UMD student and interested in joining my group, please feel free to reach out! Email me a copy of your CV, an unofficial transcript, and a short description of which projects interest you, and I’ll be In touch.
If you are a prospective MS or PhD student, please apply to the Aerospace Engineering graduate program by the December 1st priority deadline. You can find all of the requirements and information here. Feel free to send me an email letting me know that your applying to my group, so I can keep an eye out for your application.
What are the primary research interests and current projects of this group?
Our group is particularly interested in exploring the intersection of machine learning and astrodynamics research. Currently, our group investigates how we can leverage the evolving field of scientific machine learning (physics-informed neural networks, neural ODEs, deep operator nets, etc.) to enhance the quality of both dynamics models and control algorithms for flight on-board spacecraft operating in cislunar, Earth-orbiting, and small-body settings. We also are actively exploring how to expand autonomous spacecraft capabilities using advances in the reinforcement learning and decision-making communities. Broadly speaking, if it involves deep learning and satellites, we’re probably working on it.
How often do group members publish papers?
The unofficial expectation is that students should be publishing a conference paper about once a year, and a journal paper (typically building on the work of the conference paper) once every 1.5 years. Obviously research timelines can vary dramatically based on many circumstances, so this is a guideline, not a hard rule.
What’s the expected duration to complete the graduate program in this group?
It depends. If you’re joining with only a Bachelor’s degree, the median is between 4-5 years. With a Master’s degree, you may be able to transfer in course credit and have preexisting research findings which can shorten it to between 3-4 years.
How are research topics chosen? Can students propose their own projects?
Students and I will work together during the first year to develop an initial research plan. Typically this involves a blend of (1) working on a preexisting project of the lab, and (2) exploratory work based on the student’s individual research interests. The goal is to help find a topic that is sufficiently overlapping with the research themes of the group, so the student always has a strong network of resources and collaborators to further support their research goals.
What resources and facilities are available to group members (labs, equipment, software)?
We have a dedicated work space for our members on campus, and our lab has special access to high-performance computing hardware (both CPU and GPU capabilities). PhD offers to the lab typically include a new computer and various research subscription services.
How often do group meetings and one-on-one meetings with the advisor take place?
I meet with each student every two weeks for an hour. This is an opportunity to get personalized feedback and discuss research progress, administrative tasks, and learning goals. On the other weeks, we meet as a group where to discuss some topic of high-relevance to the majority of the group. Typically, one or two members will volunteer to present on high-impact papers, useful software tools, or major developments from their own research. Finally, once a year, I will give a vision seminar where I will outline the learning and research priorities for the group for that year and will ask for feedback from group members to refine it further.
Where could graduate students from this group realistically go after graduation?
All over. The research themes of our group are in high-demand, within the aerospace community and beyond. Most commonly, researchers for these topics go on to work for one of the major aerospace organizations in the country (JPL, SpaceX, Goddard, Applied Physics Lab, Blue Origin, etc); however, other opportunities include aerospace start-ups and contractors (Rocket Lab, Starfish, Omitron), machine learning research companies (DeepMind, OpenAI, Amazon), or government labs (AFRL, NRL, etc). A silly few also choose to pursue careers in academia.
Are there opportunities for interdisciplinary work or collaborations with industry?
Yes! Some of our projects are funded through industry partnerships, and students often have the opportunity to interface directly with companies and project managers associated with these partnerships. In addition, many students choose to pursue internships during the summer to better direct their research to meet real world needs (and better understand prospective employment opportunities). Moreover, our group is well connected with the CS department, so we occasionally work on cross-department grants and projects.
How does the group support students in attending conferences or workshops?
I firmly believe in having our students attend conferences and connect with our broader community. Each year, I am typically able to budget student travel for one to two conferences. In particular, our group regularly attends the AIAA/AAS Astrodynamics Specialist Conference in the summer and the AAS Space Flight Mechanics Conference in the winter. In the past, students have also attended SciTech, IAC, NeurIPS, ICML, IEEE Aerospace, AGU, and others when appropriate. Regarding internal support, the MLDS lab has internal workshops each year on presentation etiquette / expectations, and students often organize practice sessions where they can give and receive feedback from one another.
Are there opportunities for teaching or gaining teaching experience?
Yes! Graduate researchers often have the opportunity to mentor undergraduate researchers, and there are teaching assistantships where students can serve as TA or part-time instructors for a course.
How would you describe the working dynamic within the group? Is it collaborative, competitive, or something else?
Strongly collaborative. Graduate degrees can be lonely endeavors, but the MLDS Lab actively works to avoid that. We try to make sure that every new member is paired with an older member to work together on a project to ensure a smooth on-ramping process. Most of our students have one core project, and work as a co-author on one or more other projects to ensure the group is constantly exchanging knowledge and expertise. Moreover, when students write papers and fellowship proposals, it is common for others in the lab to review and provide feedback before final submission. Also, beyond purely work settings, the lab regularly gets together for meals and activities throughout the semester, so we generally all get along here.
What’s the expected weekly time commitment for research? Are there expectations for working on weekends or holidays?
Graduate degrees often attract students who are notoriously invested in their work. While that intensity can be useful, it can also play a major role in burnout. My expectation is that students put in a genuine effort during the work week, but I do not expect for students to be working during the weekends or holidays. Everyone has different paces and needs, so I try to give students the space to establish what works for them.
What’s your mentoring style as an advisor? Hands-on, hands-off, or a mix?
I aim to have a dynamic mentoring style that evolves primarily as a function of time in the lab. In the first year or two, I try to be actively engaged with you and your research. I want to make sure that I can supplement your learning, offering resources, papers, and feedback to minimize those initial growing pains. We’ll develop research plans, timelines, and experiments together, and I’ll work to cater to your personal learning preferences. Further into the program, I work to give you the space to really become the singular, world-expert on your research topic. We’ll transition to you leading the projects, timelines, and goals. At that point, I will to offer higher-level feedback about research narratives, project management strategies, and connect you with the right people for your future.
How many students do you hire each year?
It’s hard to say exactly. Each year I receive a large number of qualified applicants. In an ideal world, I would make offers to everyone. I think this work is genuinely exciting and want share the fun with as many researchers as possible. The reality, however, is more complicated, and the exact number of students that I can take on ultimately boils down to available funding. Typically I can hire 2 or 3 PhD students each Fall, possibly more if students win graduate fellowships (NSF, NDSEG, NSTGRO, etc). That said, I review each application carefully, and I am committed to hiring students with diverse backgrounds and experience. If you’re passionate about space technology and you like programming, then you’re in good company here, and I strongly encourage you to apply!
I’m interested in joining, but I don’t come from an aerospace background. Will that hurt my odds?
Not here. The majority of the aerospace content that you’ll need to succeed in my group is covered through a comprehensive set of graduate courses offered at UMD. Obviously there will be some growing pains, but I find this is true even for students who join with an AE background. Astrodynamics content is seldom covered in undergraduate AE curricula, so as long as you have your math and CS fundamentals covered you should be in good shape, and lack of explicit aerospace experience won’t reflect poorly in your application.