A conversation with Brian Glass, Principal Investigator of the Atacama Rover Astrobiology Drilling Studies (ARADS) project at NASA’s Ames Research Center in Silicon Valley. For more information on Brian’s latest field test, check out https://www.nasa.gov/feature/ames/mars-rover-tests-driving-drilling-and-detecting-life-in-chile-s-high-desert

Transcript

Matthew Buffington (Host):Welcome to NASA in Silicon Valley, episode 35. If you are in the Bay Area, we want to remind you that NASA will be at the Silicon Valley Comic Con in San Jose this weekend. Keep an eye out for several NASA booths and panel discussions, including many of the guests that have appeared on the podcast. I’ll also be walking around on the floor so don’t hesitate to say hello. Today’s guest is Brian Glass, a NASA researcher and Principal Investigator of the Atacama Rover Astrobiology Drilling Studies (ARADS) project. We talk about how his work at NASA has bridged between doing science and engineering. We also went into detail on his work conducting research in the Atacama Desert as a precursor and practice for future science missions to Mars. So, here is Brian Glass.

[Music]

Host: We always like to start it off with just learning about Brian. How did you join NASA? How did you get to Silicon Valley? Tell us about you.

Brian Glass:Okay, a little bit about myself, to start with. I got to Silicon Valley actually following in my spouse’s footsteps because she wanted to go to Stanford Law School. She had helped put me through graduate school to get my Ph.D. I was in robotics at Georgia Tech. I came from MIT to Georgia Tech. I’m an Atlanta native, originally. Came home for a bit. Finished graduate school. Was looking for someplace in aerospace. I was into large space structures, at that point, and robotics. And came here to work on the Space Station Freedom – it wasn’t Freedom then, it was just space station refrigerator. Big cooling and heating loop. And came to the automation and robotics division in its nascent form at that point.

Then went back to graduate school at Stanford in geophysics. I always had an interest in geology and geophysics and the ways the Earth morphs over time as a system, the Earth as a broad system. And then how all the patterns that we observe in how Earth changes, how those map onto other bodies, or can we find bigger picture morphological and dynamic motions and changes in evolution in landforms that happen in other places that we can learn from here?

And so if I take the two interests on the robotics and automation side and the geophysics side and put them together, it’s like, where can I find something that overlaps with those two? And so I wound up a research group leader in that same automations and robotics division where I’ve been since 1987.

In the same organization, at NASA Ames, other than a year and a half I spent in the space station office at NASA Headquarters.

Host: Growing up, were you always into NASA, wanting to do NASA stuff, or did it just happen to work out of what you were studying? It was a perfect fit?

Brian Glass:No, I was totally a child of Apollo, so to speak. 1960s, black and white CRTs televisions in the classroom showing a launch of Gemini or something in first grade, and had models and posters hanging from the ceiling when I was a kid, and the whole works, and always wanted to work here and always looked to that. Did the usual Ph.D. in your 30s, if you’re thinking about being in the astronaut corps thing. I got my scuba certification, my pilot’s license, my ham radio. And got as far as having your medical background and references checked, maybe the round of 600 a couple times, but that was it. They weren’t really looking for surface field geology or geophysics people back in the ’90s where it was really going to be more a space station focus, not planetary surface focus. I was before my time maybe in that sense.

Host: It’s like people tend to think of NASA and astronauts. But at the end of the day, it’s like they got to get there somehow. Somebody has got to build those systems. And even when they land on the moon or on Mars, it’s like what are you going to do there?

Brian Glass:Right.

Host: What is the science?

Brian Glass:What’s the field science? I guess my deployable autonomy technology group is about things that automate instruments, that get science, and things that gather samples that you put into the instruments and hopefully find out something about what you’re studying. We have another group in the same division that does rovers.

Host: Okay. Yeah.

Brian Glass:I tease Terry Fong, my colleague over there, who leads that group.

Host: Someone we’ve had on the podcast before.

Brian Glass:One you’ve had on the podcast before. His group does the X and Y axes and I do the Z, so to speak. He goes laterally on the surface and navigates, and I’m about going down into it once you get there, and what you feed with it.

Host: You said that you’d been in the same office for a number of years. How is the evolution of that been over time? What you were working on in the ’80s, ’90s versus what you’re doing now? How has that changed?

Brian Glass:I started off working on space station. Like I said, fresh Ph.D. working on AI [artificial intelligence], robotics and automating a big space station system. In the late ’80s, we had one of the first automated AI systems that actually controlled a prototype testbed. We went to chambers at Johnson Space Center. I lived there alternating weeks for about a year and a half.

And full-scale test with that. And it was successful. That actually sold the mission ops people at JSC back in that era that yes, you could actually automate these things. You didn’t have to control everything minutely from the ground. And so that was actually a little bit of an advance, a little bit of a psychological shift in how they viewed mission ops back in the late 80’s, early ’90s. Now we would take that for granted, but it was not a settled thing. You’re still coming out of the era of strip charts and watching everything and controlling from the ground. So that was an interesting thing.

Then I went on to work on SETI for taking the same principles and automating things. There was resistance there even in the radio telescope and the radio astronomy community. “Why do we need diagnostics? Why do we need automation? Why do we need anything to watch our equipment?” Until they had a fire at Arecibo in our trailer just before they were supposed to turn on the SETI targeted search at Columbus Day, big hoopla, in ’92 for Columbus’s.

Host: Yeah.

Brian Glass:But software, detecting the problem, temperatures, sounded alarm. An operator came over, said, “Oh my goodness.” Called for fire, put it out, didn’t burn up our rack of equipment. Minor repairs. SETI unveiling went off in ’92 so that it was successful, so that when congress pulled the NASA plug on it a year or two later, they were able to go private and have continued on the search since with private funding. Sometimes I will tell people that yeah, we sort of saved SETI with automation.

Host: It seems like for you, working on intelligent systems and AI stuff, but then also working on the geophysics part, it’s like you see the engineering versus science. You are able to combine those both. What’s your hypothesis, what are you trying to discover versus what are the tools you actually need to put that together. It’s like two sides.

Brian Glass:I like to write papers in both areas. I like to have one foot in the science and one foot in the engineering side. I live in that seamy realm between the two communities and how they both view one another. And I get to go back and forth and put on different hats a lot. I actually enjoy that.

Host: Yeah.

Brian Glass:I can look for, “What question are we answering with this?” The engineering is means to an end to figure out this question we want to answer from a science hat, where I can look at, “Hey, we’ve got this neat thing. This will enable us to address these, this, this and this questions from the engineering side” and find a way to try to marry these two together and point each other saying, “If you only talk to so-and-so, they have a way to do this,” or, “I think they might be interested in asking these kinds of questions if they knew you could actually help them answer it.” So it’s that back-and-forth rhythm.

There was a little bit of an intermediary point where I worked in air traffic control leading up the ’96 Olympics because I just got a pilot’s license, had just gotten back from NASA Headquarters. And my division chief said, “We need someone to work on automating air traffic management, and the aeronautics people are completely full. They can’t take on any new work and they’ve asked us to help out.”

So in a couple years, we put together a system for routing traffic on the surface and getting things more efficiently, saving a couple minutes per taxi operation, something called the “Surface Movement Advisor.” Pulling Ethernet cables up the Atlanta Hartsfield control tower over Christmas/New Year’s break. We had to wire it ourselves. This is 2017, and until 2013 or 2014, I still had property inventory on my NASA account that was routers and network equipment at Hartsfield Airport still I had to go and check on periodically. We were running their base backbone for them.

Host: Okay, so talking about the science versus the engineering and combining those, you’ve also done some work over in the Atacama Desert, some kind of analogs of what it would be like for astronauts on the surface of Mars. Talk a little bit about that.

Brian Glass:After all that background stuff, really from marrying geology, science, and the engineering side, if you want to send something, an instrument or a drill. We’ve done a lot of drills. If you want to test that, you need to put it in a relevant environment. That means putting it sometimes in a vacuum chamber and maybe shaking it, doing thermal vacuum tests. But it also means if you’re going to actually get samples and actually test an instrument, no one will believe you unless you can actually go to some relevant environment on Earth, as well as your chamber test. In terms of testing an instrument, it’s too easy, even only if subconsciously, for a human being or a group of human beings to design a test that they can pass.

Host: Yeah.

Brian Glass:“Let’s make up our constituents of this tracer, this type of spore. We’ll layer it this way,” but because they already know that’s something their instrument can find, or they know that their drill can penetrate this.

Host: So you have to build in some sort of practicality, or really protection for the unknown unknowns.

Brian Glass:Unknown unknowns. Mother Nature is suitably perverse.

Host: Yes. Nice.

Brian Glass:That’s what we need. We need not to be able to, to badly paraphrase Kennedy, not to do the things which we can predict or that are easy in our own chamber tests, but to do the things that are hard or perverse or that will surprise us. We want to be surprised. Much as I hate being surprised out in the middle of nowhere in a tent with howling winds, and then suddenly we have a problem with our software or hardware in whatever test we’re running, and my crew is saying, “Oh my goodness, we didn’t expect this.”

Host: Yeah.

Brian Glass:Much as it may be consterning in the moment, that’s a win. The fact that we hit something that we didn’t know we were going to encounter or didn’t know how to handle while we’re testing it somewhere on Earth, that means we’re finding it then and we’re not finding it after we’ve launched it somewhere. That’s a big win for us. That means we’re doing what we’re supposed to be doing. We’re supposed to be taking pieces of prototype space hardware, whether that be a drill or instrument control, take it to a difficult location and creatively try to break it.

Host: And so if you can take that to Antarctica or to the Atacama Desert in Chile.

Brian Glass:Chile. The test sites that we’ve used that my group uses the most are the Atacama, most recently, and also Haughton Crater, which is a large impact crater at about 75 [degrees] north, 90 [degrees] west in Arctic Canada. Because you have essentially breccia which is very regolith-like in its chemical and mechanical properties, and you have ice lands there. So if you want to do cold Mars analogues that involve doing something with the surface or subsurface, that’s a great place to go.

If you want to look at someplace where you’re dealing with salt crusts and evaporites and also a very barren, very little life at the surface, very dry place, the Atacama is a great place to go, sort of a hot terrestrial analogue and a cold terrestrial analogue. It’s easier to go to the Atacama in that respect, or to Haughton than it is to go to the dry valleys in Antarctica, where we’ve also tested.

Host: Okay.

Brian Glass:Which gives us sort of a cold, dry, barren… Really one of the other thrusts in the last 10 years has been to increasingly… We’ve followed the water and found the water. We’re looking for habitability and we’re finding traces or evidence of that. Next is going to be a push for biomarkers. Do we see signs of life?

Host: Molecules.

Brian Glass:Molecules. Do we see extant life? Do we see dead life? Do we see amino acids?

Host: Yeah.

Brian Glass:If we see amino acids, what’s their chirality? Are they things that we could associate with Earth-like life?

Host: As far as I’ve understood things for the most part, most of our rovers, things that have been on the surface of Mars for the most part really are just scratching that surface. I know you’ve done a lot of work on drills and stuff. So what’s the thought process going into that? Is it just a matter of getting a deeper sample size?

Brian Glass:You could argue the first, on Mars at least. We had a handheld drill that could go a meter or two with Apollo, and the Russians took a drill to the moon similarly.

Host: Okay.

Brian Glass:The moon is a difficult place for its own reasons. On Mars, we had the rock abrasion tool which just sort of scraped a millimeter or so off the top of things so we’d have a fresh surface. But you could argue that’s a millimeter down.

Host: Yeah.

Brian Glass:Curiosity has a little bitty thumb sized rotary percussive drill that go to about five centimeters for sampling outcrops, primarily. The drill bit itself is actually derived from a Hilti drill bit that you can go and buy at Home Depot that’s been machined down. But we’ve had a series of problems. And then there’s the issue with cuttings, the powder material coming off of the drill and how you screen it and sort it so that you don’t get things clogged or clogging your instruments. So they’ve had problems with that.

And so they’ve tried to mitigate as best they can and with software, and so got some good results out of it. But it’s still a fragile system and can only go about five centimeters So it’s an additional incremental step, and that’s good. But if we want to really look for things that aren’t irradiated at Mars surface, we need to be able to go deeper than that. There’s an argument, debate within the scientific community, over “how deep is deep.”

Host: Yeah, really.

Brian Glass:There’s also a decaying issue. If you’re looking for the remnants of something, biomarkers that are four billion years old, then the accumulated radiation burden and so forth that you would get, you’d see less and less that would be recognizable near the surface and you’d have to go deeper and deeper to see older and older, probably.

Host: Is that the future of where you see a lot of your work going? Is it just a matter of drilling deeper or getting bigger samples?

Brian Glass:We already have working together with our industry partners. A lot of times NASA Ames’ role in this has been not just integrating the systems, but we write a lot of the automation and AI software that controls. Because in the real world, drilling is something that humans do. It’s an art form. It’s done with roughnecks and engineers out in rugged environments, platforms and other places. You’re listening to lots of different data sources, and you can’t really see or know what’s going on a mile below the surface.

Sometimes you still have people who literally will put their hands near a shaft and feel the vibrations. They’re doing frequency domain analysis in their heads. So how would you capture those data, and frequency domain, and pattern matching things that humans are doing? Those are all interesting AI problems that we apply to diagnosing how do drills get stuck underground and how can we recover from faults when they happen.

Because if you’re sending it to Europa or to Mars or to an ocean world and you’re trying to drill down some distance, it’s going to get stuck, it’s going to encounter problems, it’s going to encounter things, like I said before, you never expected. So you have to have some ability to reason from that. And you’re far enough away from Earth that lightspeed means you can’t control it from mission ops.

Host: The huge delay.

Brian Glass:Right, the huge delay. Just as I go back full circle to that space station thermal control system I did as a fresh Ph.D., when that kind of left mission ops and mission control people in Houston saying, “We don’t have to control everything from the ground. We can do a little bit of automation.” Now we’re looking at situations where we have to do things on board. You can get a drill or a sample transfer mechanism stuck in about 5 to 10 seconds.

Host: Yeah.

Brian Glass:5 to 10 seconds. It takes 10 minutes, on average, to talk to something on Mars, and then 10 minutes to get something back. So if you depend on Earth-based control, you’re stuck by the time anything happens. That’s one reason to stick to little bitty outcropping drills like you have with Curiosity or something where if it gets stuck you can abandon it. But that’s not going to get you even a meter or two on any other body beyond the moon. Yeah, you could maybe still joystick it on the moon, but you can’t joystick it from Earth anywhere beyond that. It has to be self-contained and have some intelligence and control of its own.

Host: It seems that moving forward, it’s that teamwork. It’s like here’s some smart software. Here’s robotics. I can do it. But in working in coordination with humans, by tag-teaming it, everybody can do what they’re really good at.

Brian Glass:Right. Let the humans do the supervisory control, say, “This is what our goals are. We want to go down another 20 centimeters in this hole and then sample and then feed this instrument, that instrument. And then based on that, give us an update in six hours, the next time the deep space network gives you. And we’ll find out and figure out what to do then depending on your results and what the instruments come back with.”

So it’s that flow between humans and robots and automation and each doing what they’re doing best. The humans are doing a high-level pattern-matching and the science and the problem solving, and let the virtual roughnecks and the robotics do their thing on the surface.

Host: For folks listening who want to learn all that Brian Glass does, should they just go to NASA.gov? Probably where is the best place to go?

Brian Glass:Probably blogposts from past field deployments. There are a couple of past stories from our current Atacama robotic astrobiology project where we’re putting a two-meter drill on a mobile platform on one of Terry Fong’s rovers called KREX-2. And then life detection instruments as a payload on the same rover. We drill, we acquire a sample, we feed it. The first of those deployments with just the drill and a couple of the instruments was in February 2016, and you can find that on NASA.gov as a story.

Host: Excellent.

Brian Glass:There will be one coming up soon about the second season where we actually have a mobilized drill and sample transfer arm and three instruments in the field, one of which actually rode on the rover part of the time. Next year we will have all the instruments on the rover fed by the sampling that we’re bringing up from a meter or two below the surface of the Atacama, which we generally don’t find much at all alive, except occasionally there are layers based on the past history of the area where you’ll hit a biologically rich, say, two centimeters.

So this is a good place to test the instruments. It’s a good place to test the system. It also forces us to look at things like planetary protection and cross-contamination issues, which are going to be really important when we send something like this to look for life on Mars.

Host: Cool. So for folks who are listening, we can add those hyperlinks to those different stories. We can add those into the show notes. Also if anybody has specific questions for Brian, we are on Twitter @nasaames. We use the hashtag #NASASiliconValley. We’re just going to throw you out there. So if anybody has questions, we’ll circle back with Brian and get them out to you.

Brian Glass:Sure.

Host: Thanks for coming. This has been fun.

Brian Glass:All right, thanks for the invitation. It’s been fun.

[End]