My Conversation with Chris D'Souza
Navigation, mentorship, and NASA
Dr. Christopher D’Souza currently serves as the NASA Technical Fellow for Guidance, Navigation and Control. He’s been the Navigation Technical Discipline lead for Human Spaceflight, and Deputy Chief of the GNC Autonomous Flight Systems Branch at Johnson Space Center. He’s the principal architect of the Orion onboard navigation system, which flew successfully on Artemis I. Chris has mentored many navigators, and oversees much of NASA’s GNC development for human spaceflight.
Q: Who were your most impactful mentors? What did they teach you?
That’s an interesting question. When I started out, this idea of, “mentoring,” wasn’t a thing. Nobody had these categories for the fact that they should be speaking to the next generation and investing in it. That wasn’t something they did deliberately. So there were a few people who had an impact in my life. When I was finishing up by PhD, the fellow I was working with at the Air Force—his name was Jim Cloutier—was a mathematician by trade. He got his PhD from Rice, actually under Angelo Miele, who was one of the giants in the fields of Astrodynamics and Optimization in the 60’s and 70’s. I started interning with Jim at Eglin in the summers. He took an interest in me and we worked together for many years. He was not an engineer—his desire was to do math.
When I went to Draper, I met engineers who were phenomenal, but (at the time) they weren’t in to mentoring. They were students of Dick Battin. One of them was Tim Brand. He was one of the best engineers I’ve ever worked for. He was old school—grew up during the Apollo era. He had very high standards. There was none of this, “Hey, let’s make you a better engineer,” sort of thing.
When I came to JSC, that’s when I saw leaders and managers take an interest in the careers of younger people. That’s where I got this bug for mentoring. About 15 years ago I really took it to heart that I needed to invest in the younger generation.
Q: You wrote, with Russell Carpenter, this widely used manual called, “Navigation Filter Best Practices.” Where did the impetus for that come from?
Russell and I have known each other forever. We were at grad school together in the early 90’s. He was more into navigation than I was back in the day. I was more of a guidance and trajectory optimization person. Our paths crossed at UT (Austin), and again at NASA. He went to Goddard, and I went to JSC. When I started, I realized there was not a big navigation community here. There was lots of experience, but the tools were very immature. A part of it was that we had been flying shuttle for so long—people thought that shuttle was it.
Truth be told, I never wanted to do navigation. I wanted to do guidance and mission design. I wound up being assigned a project on covariance analysis and I developed a tool set called linear covariance analysis, which is now a workhorse at JSC and many companies. I started getting more into navigation filter design. There was a program called Restore, and I started going over to Goddard a lot. Russell and I resumed our friendship and collaborations, and so we decided to write this manual together for best practices. He had done a lot with covariance analysis in low Earth orbit. I had already been writing a lot of notes for navigation. I call them my Alzheimer's notes. It became 600-700 pages. We distilled part of that down for the manual.
Q: If you had all the R&D money you wanted, what GNC related problems would you tackle?
I’ll break it out into the three areas. For control, nonlinear control as applied to large spacecraft—big ones like Starship. The control associated with that is very challenging, including slosh. In the guidance arena, it would be being able to use some of these techniques like convex optimization and realtime optimization in order to successfully guide a vehicle either to the surface or to do Mars entry, or dock with another vehicle. Guidance has to work in different dynamic regimes—thick atmospheres, thin ones, or none—and we’ve had some great techniques in the past. Now we have better computers and can use these newer algorithms.
In the area of navigation, the area of optical navigation has been a major focus. Using distributed sensing networks for things like constellations is also a big area. There are also different types of sensors now like event-based cameras. And then, finally, how do we use machine learning in some of these areas? Once you open that can of worms, then how do you do verification and validation (V&V) on that? The more complex a GNC system is, the harder the V&V becomes. We may need to rethink how we quantify risk in a situation where machine learning is a black box? Things like obstacle avoidance is a huge issue too.
That’s just in space, but in the aeronautics arena we have urban air mobility vehicles, non-traditional vehicles, and GPS-denied environments. How do you successfully land a vehicle if a rotor or thruster goes out? How do you avoid other urban mobility vehicles? There are a host of challenges that I see, and if we had an infinite amount of money, we would spend that on all the challenges, with none being particularly more important than the other.
Q: You worked on missile systems. How did the technical aspect and the project development aspect differ from spacecraft GNC?
When I started working with the Air Force, I was working on GPS guided, un-crewed vehicles, and you can fill in boxes as to what that means. That was at the beginning of the GPS arena, when there were six satellites in orbit. They were all for military purposes, and it was an interesting time. In about five years from that the satellites became widely used for civil applications.
The whole duel-use aspect of those navigation systems are very interesting. A lot of time the military starts the process, and then eventually things get used more widely. They do a lot of stuff that sometimes we don’t know about until many years later. Sometimes, because it’s so compartmentalized, it’s hard to be innovative in those areas. It’s interesting in the military world because they tackle some very challenging problems.
Q: You’ve said that, “If you want to have job security for the rest of your life, go into navigation.” If you had an undergrad student that said they wanted to do this kind of work, what would you tell them?
I tell undergrads this, and they don’t like it so much, but I tell them you better be able to write well and present well. But with respect to navigation, I would tell them you better plan to go to grad school. It takes, on average, about 10 years to train up a great navigator from the time they begin grad school. Where they can be on their own and impact the next generation.
Navigation is interesting because you need to have a good foundation of linear systems and nonlinear systems. You need to know estimation theory, optimization theory, and astrodynamics. You need to understand general dynamics! An undergrad does not have much of that ingrained, so you have to go to grad school. And then in grad school, you need classes and work that actually addresses all of those things. We’ve got people here who have no graduate school, and they’re learning, but it will take them a while because they haven’t had the courses. And you can learn most of it, but fundamentally you still need to be able to write equations of motion and derive things.
Q: You’ve stepped into a new role at NASA of Technical Fellow for GNC. What do you hope to accomplish in this role?
I’ve thought about it a lot. One of the major things is to mentor, and include a larger variety of ages and stages of GNC engineers in the team. I’ve got a team of about 30 people that I support. I want a mix of senior level people and fresh out of grad school. That way, whenever I assemble a team to solve a problem, I can pull the fresh outs and seasoned people to impart knowledge. The other thing is to identify areas that NASA needs to grow in and invest resources into that. To bring money to bear in terms of solving some of these problems.
So it’s looking out for technologies and looking out for people. The third thing is I have to be ready to assemble a team to solve problems whenever there are problems. People from industry and academia, we are a community. They can differ in their opinions on how to solve a problem, but in a final analysis we ought to be able to respect one another and pull one another along.
I am hoping to leave a community behind when I leave this job.
That is all for my conversation with Chris D’Souza.
Thanks for reading,
-William Fife