Dec 7, 2018
My guest this week is Mark Micire, group lead for the Intelligent Robotics Group at NASA’s Ames Research Center. Previously Mark was a program manager at DARPA, an entrepreneur, and a volunteer firefighter.
The topic of this conversation is how DARPA works and why it’s effective at generating game-changing technologies, the Intelligent Robotics Group at NASA, and developing Robotics and technology in high-stakes scenarios.
Links
DARPA Defense Sciences Office
First DARPA Grand Challenge Footage - looks like a blooper reel
Transcript
Ben: [00:00:00] [00:00:00] Mark, welcome to the show. I actually want to start let's start by talking about the campfire.
[00:00:04]Camp Fire
[00:00:04] So we have a unprecedented campfire going on right now. It's basically being fought primarily with people. I know you have a lot of experience dealing with natural disasters and Robotics for emergency situations. So I guess the big question is why don't we have more robots fighting the campfire right now?
[00:00:26] Mark: [00:00:26] Well, so the believe it or not. There are a lot of efforts happening right now to bring robotics to bear on those kinds of problems. Menlo Park fire especially has one of the nation's leading. Groups, it's a small called kind of like a squad of folks that are actually on Menlo Park fire trained in their absolute career firefighters who are now learning how to leverage in their case.
[00:00:57] They're [00:01:00] using a lot of uavs to to do Arrow aerial reconnaissance. It's been used on multiple disasters the we had the damn breakage up in almost the same area as campfire. And they were using the the uavs to do reconnaissance for for those kind of things. So so the the ability for fire rescue to begin adopting these two new technologies is always slow the inroads that I have seen in the last say five years is that they like that it has cameras.
[00:01:32] They like that it can get overhead and can give them a view they wouldn't have been able to see otherwise the fact that now you can get these uavs. That have thermal imaging cameras is frighteningly useful, especially for structure fires. So that's so that's the baby steps that we've taken where we haven't gone yet that I'm hopeful we'll eventually see is the idea that you actually have some of [00:02:00] these robots deploying suppressant.
[00:02:01] So the idea that they are helping to, you know, provide water and to help put out the fire that that's a long leap from where we are right now, but I would absolutely see that being within the realm of the possible. Sybil about gosh now friend 2008. So about 10 years ago NASA was leveraging a predator be that it had with some with some.
[00:02:27] Imagery technology that was up underneath it. Um to help with the fire that was down in Big Sur and I helped with with that a little bit while I was back then I was just an intern here at Nasa and that's I think a really really good example of us using of the fire service leveraging larger government facilities and capabilities to use Robotics and usually these and other things in a way that the fire service itself frankly doesn't have the budget or R&D [00:03:00] resources to really do on their own.
[00:03:00]Ben: [00:03:00]
[00:03:00]So you think it's primarily a resources thing
[00:00:00] Mark: [00:00:00] t it's a couple factors there's resources. So, you know outside of I'll say really outside of DHS. So the problem that homeland security has a science and technology division that does some technology development outside of that. There's not a whole lot of organizations outside of commercial entities that are doing R&D a for fire rescue the it just doesn't exist.
[00:00:28] So that's so that's that's your first problem. The second problem is culturally the fire service is just very slow to adopt new technology. And that's not it. It's one part. You know, well, my daddy didn't need it in my daddy's daddy didn't need it. So why the heck do I need it right at that?
[00:00:49] That's it's easy to blame it on that. What I guess I've learned over [00:04:00] time and after working within the fire service is that everything is life-critical? There's very few things that you're doing when you're in the field providing that service in this case Wildfire response where lives don't. Kind of hang in the balance.
[00:01:09] And so the technologies that you bring to bear have to be proven because what you don't want to do is bring half-baked ideas or half-baked Technologies and frankly have your normal operations have have that technology in a fail in a way that your normal operations would have provided the right kind of service to protect those lives God.
[00:01:33] So the evaluation and also kind of the acceptance criteria. For technology is much much higher in especially the fire service. Then the many other domains that I've worked in. I can only think of a few other ones and you know, like aircraft safety and automobile safety tend to be the same where [00:05:00] they're just very slow to roll in Technologies and other things like that, but those two areas that I just described have government and other groups that are providing R&D budgets to help push that technology forward.
[00:02:06] So when you get the. You get the the combination of we don't have a lot of budget for R&D and we're very slow to accept new technology because we have to be risk adverse that those two tend to just make that domain of very slow-moving Target for new technologies.
[00:02:21]Enabling Innovations in Risk Averse Domains [00:02:21]
[00:02:21]Ben: [00:02:21] that actually strikes me as very similar to to NASA.
[00:00:03] Actually. We're , there's always the the saying that you know, you can't fly it until you've flown it and do you see any ways for. Making Innovations happen faster in these risk-averse domains you have any thoughts about that?
[00:00:16]Mark: [00:00:16] It's it's tough. I mean so short short answer is I don't know. I've been trying for the last 15 years and [00:06:00] I'm still still swinging at it the.
[00:00:29] The trick is just to keep going and ultimately I think it just comes down to exposure and the folks the the decision makers within the respective Fields just being comfortable with the technology. So as we now have automobiles that are sharing the highways with us that are controlling themselves and I'm not even talking like fully autonomous, you know, driverless Vehicles, you know, the fact that we have, you know, Tesla's and other high-end cars.
[00:00:59] They have Auto Pilots that are Auto steering and Lane keeping and stuff like that the ability for folks within the fire rescue domain to start becoming comfortable with the idea that machines can make decisions that are in life. Critical scenarios and if they can make the right decision on a regular basis, it sounds weird to say that something completely removed from the fire service may help improve the ability for fire service to adopt those [00:07:00] Technologies.
[00:01:27] It seems weird to think that that's the case. It's absolutely the case and I you know, like I've been doing this for longer than well. I guess 10 or 15 years now as much as I hate to admit that and I've seen a dramatic change in that now I can go into a room and I can talk about. Averaging and unmanned air vehicle and I'm not laughed out of the room.
[00:01:48] There's a comfortableness now that I see these domains accepting that they wouldn't before so, you know, hopefully we're making inroads. It's not going to be a fast path by any stretch. Yeah culturation is something that I don't think people think about very much when it comes to technology, but that's a really good point.
[00:02:09] I have geek we don't and that's that's unfortunate. And the one thing I've learned over time. That as Geeks we have to realize that sometimes the technology isn't first that there's a lot of other factors that play in.
[00:02:20]Mark's Mission [00:02:20]
[00:02:20] Ben: [00:02:20] Yeah, absolutely. something that I want people to hear about is I feel like you're one of the most [00:08:00] mission-driven people that I know and not to put you on the spot too much but could you tell folks what you do?
[00:00:07] Like why you do what you do?
[00:00:11] Mark: [00:00:11] Um well and it really depends. I'll say in yeah, you can appreciate this a depends on what it is. I'm doing so, you know for my day job. I work at Nasa have always been a space geek and an advocate for humans finding ways of working in space and one of the best ways that I have found that at least for my talents that I can help enable that is to leverage machines to do a lot of the precursor.
[00:00:42] Work that allows us to put humans in those places. It turns out strangely enough of it a lot of the talents that I use for my day job here also help with work that I do on the side related to my role as search and rescue Personnel in FEMA [00:09:00] that a lot of the life safety critical things that we have to do to keep humans alive in the vacuum of space also apply to.
[00:01:11] Women's Safe and finding humans at and around and after disasters and so I've always had this strange kind of bent for trying to find a technology that not only ties to a mission but then you can very clearly kind of Point your finger at that and say well that's that's really going to help someone stay safer or do their job more effectively if they had that piece of equipment.
[00:01:39] Those are fun, you know. An engineering standpoint. Those are the kind of Base requirements that you want and and it always helps with there's a lot of other technology areas that I could have played in and I like the fact that when I'm when I'm making a design decision or an engineering trade that I can look at it and really grounded out [00:10:00] into okay.
[00:02:02] Is that going to make that person safer? Is that going to make them do their job better? And it's really motivating to be able to. To have those as kind of your level one requirements as you as you try to design things that make the world better.
[00:02:14]
Intro to IRG [00:02:14]
[00:02:14]Ben: [00:02:14] and So currently you're the head of IRG.
[00:00:05] Yeah group lead is the official title. So I'm the group leader of the intelligent robotics group. Yeah, and I bet that many people haven't actually heard of the intelligent robotics for group at Ames which is kind of sad, but could you tell us a publicly shareable story that really captures IRG as an organization?
[00:00:22]Mark: [00:00:22]
[00:00:22] Serve, yeah, well, I would say that it is it is a an interesting Motley Crew of capabilities that that allow robots to go do things and all kinds of different domains. We have folks within our group. That specialize in ground robotic. So we have [00:11:00] Rovers that have quite literally gone to the ends of the Earth and that we've had them up in the northern Arctic.
[00:00:49] We've had them in desert in Chile. We they've roamed around just about every crater or interesting Landmark that we have in California here and long story short. We have folks that not only work with and make ground robotics smart, but then. Of them and one of the things I adore about the team is that they're all filled capable.
[00:01:13] So we all subscribe to the philosophy that if we're not taking this equipment out in the field and breaking it. We're probably not learning the right things. And so none of our robots are garage queens and stay inside inside of the lab that we love like to take our stuff outside and take them out into these domains where they're really really tested.
[00:01:34] We have a group here. Subgroup within RG that's working on Technologies for the International Space [00:12:00] Station. So we have a free flyer and have worked with many of the free Flyers that are up on the International Space Station. Now, there's a new one that we are building. That should fly very soon here called Astro B, which is all you can think of it as in astronauts assistant.
[00:02:01] So it's able to not only do things on its own but hopefully will be helpful to astronauts and also allow ground controllers to to have a virtual presence on the International Space Station in the way that they the way they haven't been able to. Let's see. We're turns out that when you're working with robots like this having very good maps and representations of the world's that you are exploring becomes important.
[00:02:27] And so we have a sub grouped here. That works with planetary mapping. So in the best, I guess most digestible way of describing that is that if you've ever opened up [00:13:00] Google Google Earth and kicked it into Google moon or Google Mars mode. That most of the especially the base in Ministry imagery and other products that are in that in that Google Earth We're actually generated by my group.
[00:03:00] And so it turns out that when you get these pictures and imagery from satellites, they're not always right and they need a lot of kind of carrying and coercing to make them actually look correct. And so my group has a suite of software. Where that's all publicly available the that can be used to make that imagery more correct and more digestible by things like Google Earth and other systems like that and then you know in general we at any given time have, you know, north of 30 to 40 researchers that are here.
[00:03:38] Doing all kinds of work that is relevant to robotics relative [00:14:00] relevant to space and yeah, and it's an awesome group and every single one of them is motivated and exactly the right kind of ways.
[00:03:52]Organizational Nitty-Gritties: IRG [00:03:52]
[00:03:52] Ben: [00:03:52] Yeah. I mean having having worked there I completely agree with that statement from personal experience.
[00:03:58] And actually related to to the motivations something that I really like doing is digging into the nitty gritties of organizations that really generate Innovations. So so look what tell me about the incentives that are at play in IRG like what really what motivates people like, how are people sort of rewarded for success and failure and how do those pieces work?
[00:00:12] Mark: [00:00:12] Well, I and. I'm going to say this and it's going to sound super simple. But the IRG is one of the few places and it's one of the reasons why I wanted to when I was given the opportunity to be the group lead that I took it is I still feel like I RG is like one of the last one of the few places. [00:15:00] I guess I'll say where the research can kind of be up front.
[00:00:34] We're creativity can be king and we can kind of focus on doing the good work in a way that I'll just say that is a little bit more difficult when you're out. A commercial world because you know chasing the next product sometimes has a whole bunch of things that come along with it. You know, what is the market doing what you know our is this going to be supported by Senior Management other things like that that we that we don't have to deal with that as much it has to align with NASA's Mission.
[00:01:06] It has to align with what the focus is our of the agency, but I will. That because we have such good researchers here our ability to create a proposal. So we end up just like everyone else writing proposals to to NASA itself and winning those proposals that they that they were kind of ization is actually in the [00:16:00] fact that these researchers get to do the research that they're wanting to do and all the research that's being handed down to them by, you know, a marketing team or some corporate exec.
[00:01:40] The other thing that is huge here and I know. Probably experienced it during your tenure when I say the folks are here for the right reasons. We all know every single person within IRG and I'll say that within especially NASA Ames out here in Silicon Valley every single one of us could go a thousand yards outside of the fence and be making two to three times what we make working for the government.
[00:02:08] And that's not it's not so much a point of Pride. But what it does is it just helps relieve the the idea that folks are are are here for the money you're here for the research and you here for the science. I use the best analogy I make quite often is I used to. [00:17:00] I used to teach as an Adjunct professor at a community college doing and this is more than this is about 15 years ago in the courses were on like PC repair and other things it was this certification called A Plus and the I used to confound the other professors because I used to always take they had one section that they would do and it was 8 a.m.
[00:02:52] On Saturday morning. It was like a it was like an 8 a.m. To 1 p.m. And it was just one day a week and I used to always take that one and the other professors were like, why are you taking an 8 a.m. Saturday course and I would smile at them and say. Because every single student it's in there. I know they want to be there.
[00:03:14] I know that they are motivated and want to be there because no one in their right mind is other than a motivated student is going to get up at 8 a.m. On a Saturday morning to go learn about PC repair and to add in to everyone's surprise, but not my surprise. I had a [00:18:00] 100 percent pass rate on that test because it was independently tested outside out of out of the classroom and I.
[00:03:39] So just smile because it was like wow, you must be a great professor and I'm like, no, I've got great students because they all are motivated to be there. So that's effectively what I have here within NASA sitting inside of you know, this Silicon Valley bubble is I have a whole bunch of frightening Lee smart people that are motivated to do good science and have absolute have financial reasons to go elsewhere and decided for themselves.
[00:04:07] This is where they'd rather work. Yeah and do so the in terms of the majority. Let's break that down a little bit the way that projects happen is that you do a proposal to like, who do you propose projects to I guess [00:19:00] is the the correct question. Well the the fun part and this is one of the the freedoms and NASA has.
[00:04:36] Can really propose to anybody we have projects here that our commercial so we work with like for instance. We're doing work with Nissan on autonomous vehicles. And and if actually done some really really interesting work there, you know related to visualization and other things like that which which borrows a lot from work that we do with the Rovers so so we can work with companies.
[00:05:03] We work with Within. First so NASA itself. One of the ways that NASA works is that because we have multiple centers, you know, NASA Ames for instance in our group will propose to NASA headquarters. So we just pitched a couple of months ago we pitch to a program that was doing satellite-based Technologies and I flew to NASA headquarters in DC and we [00:20:00] pitched it to a much like you would do to a VC or any.
[00:05:35] No any funding source, if you were a company doing it in the valley and you pitch it and we and we want it. We also work with other government agencies. So we have done work for DARPA. We've done work with the Marine Corps. It turns out that the dod Department of Defense is interested in a lot of the ways that we have worked with autonomous vehicles as the Department of Defense tries to figure out how they want to work with autonomous vehicles.
[00:06:05] So it's easy for us to open a conversation with Department of Defense and say hey, here's what we did for our Rovers our uavs or whatever and this may be something that you know, you may want to consider and a lot of times they'll come back and say well look we not only want to consider that but we'd also like to go ahead and kind of put you on proverbial payroll here and how do you either do the work for us or help us [00:21:00] understand?
[00:06:30] You know, what are the important parts of this we can work with Academia? And so we will often have projects where we partner with a university and we will go in and do a joint proposal either to NASA or all of the different funding sources that that are out there. And so NASA. Has a lot of flexibility in a way that you know myself having previously worked in the Department of Defense.
[00:06:58] NASA can do something unique and that NASA can be a consult or NASA can do work for a private company. We have a thing called a space act agreement and like the Nissan workout was talking about there. It seems odd that that a government organization would be able to receive a paycheck if you will.
[00:07:18] Yeah from a private Corporation. And it turns out that NASA has a very unique way of doing that and we leverage that frankly as often as we can. [00:22:00] So I realized that's probably a really long answer to a simple question and that's to say we can take money from just about anybody as long as it is legal and it benefits NASA in some way.
[00:07:41] Those are the only two real catches that we have. We You Know It ultimately has to benefit and NASA's Mission as. You know being Shepherds of taxpayer dollars, but as long as we can justify that we can work with a lot of different funding sources.
[00:07:58]Aligning with NASA's Mission [00:07:58]
[00:07:58] Ben: [00:07:58] And what is NASA's mission right now? Like how do you know whether something is within the purview of NASA's Mission or not?
[00:08:08] Mark: [00:08:08] Well, I NASA takes its guidance from from a lot of different places. I mean, we you know, there's the two A's. NASA, you know with respect to you know Aeronautics. I'm sorry, the what we have Aeronautics and we have space right and those are the two kind of built into the name, you know missions that are in there.
[00:08:29] We [00:23:00] also you know, the we take direction from NASA headquarters. And they are putting out, you know, we have the science side, especially for space which is really driven a lot by the Decay deal surveys and other kind of Direction with respect to where we want to see and it sounds kind of funny to say but it's like where we want to see mankind go in terms of, you know, space exploration and other things like that, but we also have Earth Sciences, you know, some of the kind of flipping back to to the the.
[00:09:02] It's up in Northern California some of the some of the best especially satellite imagery that is coming through there's actually being processed through NASA's Earth Sciences missions. And so, you know, there's a worldview and a bunch of other tools that are out there that as as the Earth Sciences.
[00:09:24] With all of the different things that are affecting especially, you know, the climate and everything else. It turns out that [00:24:00] NASA's mission is also to benefit that and to help with Earth observations in a way that ultimately helps us understand how we might be impacting other worlds when were able to achieve going there
[00:09:42]
[00:09:42]NASA-> DARPA [00:09:42]
[00:09:42] Ben: [00:09:42] Got it. I'm going to transition a little bit from your time at Nasa to then your time at DARPA. And what I wanted to know is like what were some of the biggest shocks transitioning from NASA to DARPA and then now back from DARPA to NASA because they're both government agencies, but they feel like they have very different fields at least from the outside.
[00:00:20] Mark: [00:00:20] Yeah. Um, gosh, that's there's especially from NASA to DARPA. It was I guess the biggest things that come to mind one as a program manager. It is frightening Lee empowering to go to an organization where you know [00:25:00] where you're at Nasa here. We you know with Ed My Level and with the group kind of scenario that I just described to you.
[00:00:51] We're in the trenches right? We're trying to do the science. We're doing the research and we're we're trying to make a kind of an impact at a kind of a ground level right when you go in as a program manager at DARPA your your. Trying to change a field. So you have your basically being given the power to say within this field within this field of let's say autonomous vehicles.
[00:01:19] I see the following Gap and in stating that and in creating kind of the the request for proposals and other things that you do that bring researchers to darpa's door you're saying. You're not saying I'm going to go do this technology technological thing you're saying I think everyone needs to focus on this part of the [00:26:00] technology landscape.
[00:01:44] That's a that's a different conversation at a very different level and it was startling to be frankly one of those program managers where you say. Hey, I don't think the field is doing this right and then to have an entire field turn to you and say oh, okay. Well then let's. From the thing that you want that you're suggesting that that's that isn't interesting and kind of empowering position to be in.
[00:02:11] but has a NASA does too but DARPA specifically especially with Department of Defense type technologies that eventually roll out into civilian use your ability to just speak at such a different level and at a level that is. Accepting of risk in a way that NASA does not do that for DARPA. You almost have to have if it's not ready [00:27:00] yet.
[00:02:43] If it's not risky enough that you can have a program not basically make the cut DARPA because it didn't have enough custo. It didn't have they call it and dark within DARPA. They called The Laugh ability test and that if your if your idea isn't just crazy enough that it's almost laughable. Then then then it didn't it didn't it's going to have to work a lot harder to get there.
[00:03:07] And so I'd say the probably I guess in conclusion the risk and just the empowerment to move an entire field than a different Vector that that would probably be the biggest difference as I had between between my NASA world and then going over and being able to Moonlight as a program manager
[00:03:26]Fields Impacted by DARPA [00:03:26]
[00:00:00] Mark: [00:00:00] and what are some fields that you. All like DARPA has really moved that concept is incredible and makes sense. And I it hasn't been expressed. So concisely before I'd love some [00:28:00] examples of that.
[00:00:02] Mark: [00:00:02] What are the best and I think the most recent example that we can now see the impact for is is autonomous vehicles.
[00:00:12] I mean you have to remember the that that now is over a decade ago that the original that the first DARPA Grand Challenge happened and what you know, I was reflecting on this while I was being chased down by a Tesla on the way into work this morning that clearly was autonomously driving itself. And I remembered that in most people forget that the first arpa Grand Challenge.
[00:00:38] First of all was millions and millions of dollars in investment and no one won. Yeah one got to the finish line. And in thinking about risk and thinking about risk acceptance what I think that's one of the best data or a really good data point of darpa's not only saying this is really hard. We're going to call it a Grand Challenge and we're going to have these [00:29:00] vehicles basically racing across the desert that if that wasn't gutsy enough from a risk standpoint, but they also then failed and then did it again and said, you know what week we literally had.
[00:01:16] Humvee flipped over on fire on in the desert and that was on the evening news for everyone to enjoy to the embarrassment of DARPA and the dod and everybody else and then they said you know, what? No, we're going to double down. This is really worth it. And we need to make this happen and the the impact for that is huge because that then became, you know kind of the ground floor.
[00:01:46] Of the vehicles that we now have running around especially out here and you know in the Bay Area you got fully autonomous vehicles now that are able to navigate their way through, you know through all of the different difficulties in the complex situations [00:30:00] that can be presented. The folks that were that Noble Sebastian threatened and his Stanford team that won the the the Grand Challenge that those people went on to to work for you know, what was the Google autonomous car which eventually became way Mo and all of the different companies and talented is sprung out of all of that.
[00:02:25] That was all born over a decade ago by an organization that is using your taxpayer dollars to do. Risky things and to say for this it's for this autonomy thing. We really think that vehicles are where the money needs to be spent and spent in a real way that that takes guts and it's still in my mind one of the only organizations that really able to kind of make an impact like that until that entire field.
[00:02:53] Hey, I don't think you're doing this right and here's what I want you to do and I'm going to put money behind those words and we're going to go change the world and a [00:31:00] decade later. We've got autonomous vehicles quite literally beside you on the highway. That's pretty awesome.
[00:03:07]Levels of Risk DARPA Shoots For [00:03:07]
[00:03:07] Ben: [00:03:07] That is incredibly awesome.
[00:03:09] Do you have a sense of what the level of risk that you're shooting for is I'm thinking just sort of. Like what is the the acceptable or even desired failure rate or is there a sense of how many fields per decade you're shooting for? Right because you think about it and even if it's changing one field per decade.
[00:03:42] The amount of change that comes out of something like autonomous cars or even the human computer interaction that came out came from the 60s might even make the whole thing worth it. So do does anybody even think about it in terms of numbers at all?
[00:32:00] [00:00:03] Mark: [00:00:03] So I never heard it framed that way the thing that the Mantra that was always drilled into US was that that it was that the way that you kept score was by the number of Transitions and how how DARPA and I guess that's more of a general DOD term.
[00:00:25] That's to say for something you create how many times did. Someone take that technology and go use it for something and so, you know, we would count a transition as you know, well, you know Army decided to take our autonomous vehicle and use it for this but we also got contacted by Bosh and they are interested in leveraging that thing that we built with this new sensor that they're commercially making available and we provided the missing link that now allows them to use that safely.
[00:00:59] Vehicles and so you kind of keep score internally on [00:33:00] that basis. The other thing though that darpa's doing is you got multiple horses in the race. So DARPA is organized into multiple floors that have different specializations. So they have like and just a couple examples. I have like the biological technology office and the micro technology office and each one of those.
[00:01:29] Floors has a specialization in so the idea that you're bringing in these program managers, you're empowering them to go change their respective fields. And then you're doing that across multiple broad domains like biology and micro technology and other things like that. That's pretty that's pretty and that's awesome in a way that it provides overlap because when I was for instance where I work, What's called DSL which is the defense Sciences office, which is to say it works on [00:34:00] kind of first principle science and physics and Mathematics and other things like that the fact that you can as somebody who's working that go talk to somebody who was fundamental in the development of mems technology, which is what MTO the micro technology office.
[00:02:21] That's what they work. And then you want to see how let's say that new chip that is leveraging mems technology might. By law might be able to parallel or be inspired by biology and go get one of the experts from the biotechnology office to you know to scrimmage on some new idea that you're having or whatever that that's that's awesome.
[00:02:44] And what that does is that just ends up being kind of this this this multiplier this Catalyst for innovations that are then, you know, you've got multiple domains that are all kind of being affected in the same kind of positive feedback loop. So I would say that's the biggest thing to directly to your question that I don't ever remember anybody saying, okay.
[00:03:03] We're not [00:35:00] hitting quote. We need you know, we need another six domain changing ideas organize, you know not have satisfied or obligation of Congress. I don't ever remember any kind of conversations like that.
[00:03:16]Organizational Nitty Gritties: DARPA [00:03:16]
[00:03:16]Ben: [00:03:16] Yeah that description of the like cross-disciplinary interactions is shockingly similar to some descriptions that I've heard of bell labs and it's the parallels that are really interesting.
[00:03:32] And I want to dig into sort of the organizational nitty-gritties of DARPA as well. So all of the the program managers who are the sort of the the drivers of DARPA, you're all basically temporary employees. And so how did the incentives their work? what are your goals as. Program manager and what drives people, what incentivizes them to do their work?
[00:00:04]Mark: [00:00:04] [00:36:00] Well, so you're right you're there. So as a DARPA program manager you therefore it's. Typically to two-year renewable contracts. So you you go in you have basically two years at which point you're evaluated as to how well your programs are doing and then you may be renewed for typically another two years.
[00:00:26] Most program managers are there for about three years and that that's kind of the the center of the bell curve is three years the motivation simple and that you're you're being. Given one of the largest. within certainly within DOD. If not within just the overall research community and DARPA has a bit of a Swagger.
[00:00:51] It has a bit of a like a brand recognition that when DARPA says that we are going to now going to focus on this particular type of sensor this particular type of technology that you as [00:37:00] a program manager. You have the ability to go talk to the best of the best the the the folks that are. Either changing or moving or working in those those respective technology bases that you can drop somebody an email and the fact that it's you at DARPA dot mil that that will probably get you a response that you might not have been able to get otherwise.
[00:01:28] And so so that's you know, that's I would say one of the biggest kind of motivators that are incoming program manager has as they're going in and then the the other big motivator there is you're not you're there for a limited amount of time. So for years may sound like a lot of time it's not it's really is not because you it takes about to go from like idea on the back of a napkin.
[00:01:57] To you know to kick off of a program it takes about a year. [00:38:00] There's a for as much as it looks like it's loose and free and a little crazy in terms of the ideas and stuff like that. It turns out that there's a pretty regimented all jokingly call it a hazing ritual that's on the backside that involves multiple pitches.
[00:02:21] There's a level of. Programmatic oversight called a tech Council that you have to go present to that is extremely critical of whatever it is. It is that you're you're presenting and I'll admit it some of the toughest pitches and certainly the toughest like presentations that I ever prepared for. My first tech Council was way more difficult than anything I ever did like for my PhD dissertation or anything.
[00:02:52] Like that and so yeah, and so, you know once the so if if you're on a let's say a three year time scale and it takes you a year to get a [00:39:00] program up and running you have got enough time to maybe make two or three dents in the universe, which is what you're hoping to probably do when you go in the door.
[00:03:16] And then the other thing that could happen is as program managers are cycling out. So, you know you everybody's on kind of disorder. Even in their out after three years the other program managers have to then inherit the programs that are run up and running that some previous program manager, you know may have pitched in awarded but is now headed off to you know, make you know, buku bucks and industry or whatever and so it's another disc I'll say distraction that you have because program managers sometimes naively myself included go in thinking.
[00:03:47] Okay. Well, I'm just going to go in and. Ditch my own ideas, and I don't even know what this inheriting other programs thing is but I'm going to try to avoid that as much as possible and now you've got three or four or five different programs that you're running and hopefully what you've done is you've built a good [00:40:00] staff because you're able to assemble your own staff and you can kind of keep keep the ball running but that's kind of a that's the cycle if I can give you kind of a you know, the the the day in the life kind of you is that you're going to go in.
[00:04:19] You're going to be pitching and coming up with new ideas and trying to get them through Tech Council. Once they get through Tech Council, then you've got a program up and running in as soon as that programs up and running then you've got to be looking toward the next program while your staff. You know the ball rolling on your other on your other programs, then you rinse and repeat at least three or four times
[00:04:43]What does success or failure look like at DARPA [00:04:43]
[00:00:00] Ben: [00:00:00] and what does the end of a program look like either success failure or question?
[00:00:11] Mark: [00:00:11] Um, it depends on the program and it depends on the objectives of the program, I guess, you know, the grand challenges always end with [00:41:00] a huge Fanfare and robots presumably in a running through Finish Lines and other things like that. There's other programs that end much much more quietly. We're a technology may have been built that is just dramatic.
[00:00:37] We enabling and and the final tests occur and a lot of times DARPA may or may not have an immediate use for the technology. Are that the reasons for the Technology Building being built. Innocence the program started and so you may see the companies basically take that technology back and continuing improving on it or incorporating it into their products and you know, and that's a very kind of quiet.
[00:01:07] Quiet closure to what was a really really good runner really really good program and then presumably you would see that technology pop up and you know in the consumer world or in the, you [00:42:00] know our kind of our real world, you know in the next four to five years or so, and so there's a it's the full spectrum as you would probably imagine that that some of the program's some of them fail loudly some of them fail quietly.
[00:01:35] And the successes are the same some of the successes are with great Fanfare and then other times and I'll say some of the most enabling Technologies are out there sometimes close their their time and their tenure at DARPA very quietly. And then some years later go on to do great things for the public.
[00:01:53]How DARPA Innovations get into the world [00:01:53]
[00:01:53] Ben: [00:01:53] That's something that I hadn't thought about so the sort of expectation of the model of how the the technology then gets. Into the world is just that the people who are working on it as part of the program are then the ones to go and take the ball and run with it. Is that accurate?
[00:02:18] Mark: [00:02:18] Absolutely, and [00:43:00] I'd say that that's a difference so strictly speaking.
[00:02:22] No research happens Within darpa's Walls, and I guess that's one of the things that that both Hollywood and the description of DARPA. Sometimes get confused is be. That you know DARPA is this this, you know, presumably the warehouse full of mad scientists and you go inside and everybody's in lab coats and it looks like something out of X-Files and that's not it's not the case at all that DARPA is there to to first, you know catalyzed Technologies for DOD purposes, but.
[00:02:59] But those those folks that are working for DOD are also companies that are producing products made many of them are producing products that are very much outside of DOD and so the spillover and the fact that the DARPA can and I'll say relatively quietly create technology that [00:44:00] is that is just it's a catalyst for the greater good or the the greater use of Technology more broadly that that is a it's a wonderful.
[00:03:28] Ability that DARPA has that a lot of other labs don't have that ability to do so you take and I'll give you an example. So let's take like either Air Force research Labs or Army research lab or any of the research Labs that are with the particular branches of the military that does have actual researchers much like NASA Ames here.
[00:03:49] We have actual researchers that are inside of our four walls that are doing work and we can do work that you know is it can be exclusive to the government? But but in darpa's case because there is no research being done within its four walls that most of the contractors most of the what they would call the performers the folks that are performing the technology development that depending on the mostly depending on the contract and the contracts are usually written such that those companies can take those Technologies and and use them for [00:45:00] whatever they'd like after the the terms of the contract is done
[00:04:26]Improving the Process of Getting DARPA Innovations into the world [00:04:26]
[00:04:26]Ben: [00:04:26] something that I've always wondered is you try so many things at DARPA and there's there's no good way of sort of knowing all the things that have been tried and what the results were. Is there any there ever any thought. having it better knowledge base of what's been tried who tried it and what the result was because it feels like for every technology that was developed by a company who then picks it up and runs with it.
[00:00:04] Sometimes there's a something that's developed by a lab that. Is full of folks who just wanted to do the research and sort of have no desire to then push it out into the world so is there is there any effort to make that better make that [00:46:00] process better?
[00:00:06] Mark: [00:00:06] Yes, and no but this is a bit of a trick question and I'll answer that.
[00:00:12] Well, I'll answer the tricky part. First of all, let me ask let me back up. The obvious answer is that DARPA especially within the last five years or so on his been working much harder to be more open with the public about the work that's being done. You can hit darpa's website and. To the 80th percentile of an understanding of the work that's being done within within DARPA did that the balance of the twenty percent is stuff.
[00:00:44] That's either classified or is of a nature where you would just need to do a little more digging or talking with the program manager to really understand what's happening. Okay. So that's a straightforward answer the trick. The trick answer here is that it's better sometimes. Have folks go in that don't know their [00:47:00] history.
[00:01:05] The don't know why that previous program failed because since that previous program ran technology may have changed. There may be something that's different today that didn't exist 10 years ago. When that was when that program was also tried the there was this interesting effect within DARPA that because your.
[00:01:31] Managers out about every three to four years and because I'll say it like this because DARPA didn't in the past had not done a very good job of documenting all of the programs that it had been running that there was a tendency for a program manager to come to the same Epiphany that a equivalent program integer had come up with a decade earlier.
[00:01:56] But that doesn't mean that that program shouldn't be funded. Now. There were folks within DARPA that had been there for a long [00:48:00] time. So interestingly enough the caught the support contractors, so we call him sita's which is systems engineering technical assistance, and there are some CDs support staff that has been there for multiple decades.
[00:02:20] So they were back at DARPA during the you know, roaring 80s and 90s, which is kind of, you know, some of the the Heyday for some of the more crazy DARPA stuff that was happening that you would have a program manager go and Pitch some idea. Timers in the back start, you know lean one would lean over to the other one in elbow on the you know, and pointed a slide and they both Giggle and then you would ask them later is like hey, what was the what was the weird body language?
[00:02:48] He's like, yeah, you know, we tried this back in the 90s and and he didn't work out because Laser Technology was insufficiently precise in terms of its timing or you know, some other technical aspect or whatever, but it's good to see you doing this because I think it [00:49:00] actually has got a fight.
[00:03:06] Chance of making it through this time and hearing that and watching that happen multiple times was interesting because we tend to We tend to say oh well if somebody already tried it and I you know, I'm probably not going to try it again. Whereas with DARPA that's built into the model. The the the ignorance is an essay.
[00:03:26] It is ignorance. It's not necessarily it's ignorance of the fact that the idea and the Epiphany you just came up with may have been done before. For that is all I want to believe it's by design that then they will allow a program to be funded that may have been very similar to one that was funded earlier.
[00:03:48] But because it's under a new it's under a whole new set of capabilities in terms of technology that if you do that intelligently that that's actually a blessing for for folks that are trying to come up with new programs.
[00:04:04]The Heilmeier Catechism [00:04:04]
[00:04:04] Ben: [00:04:04] The [00:50:00] concept of forgetting things that has been tried feels almost Blasphemous in the the face bright.
[00:04:12] It's right. Like that's why I do wonder if there's sort of a middle ground where you say we tried this it failed for these reasons and then whenever someone wants to pick it up again, they can they can know that it's been tried and they have to make the argument of this is why the world is different now.
[00:04:31]Mark: [00:04:31] yes. So that is actually part of within DARPA and one of the framings that they use for pitch is this thing called a Heilmeier catechism and and it's basically a framework that one of the previous DARPA directors made that said if you're going to pitch an idea pitch it within this Framing and that kind of helped that will help you kind of codify your argument and make it succinct one of the Lines within the.
[00:00:27] Ism is why is this why can this happen now and that addresses that [00:51:00] kind of ignorance that I was talking about before as a program manager when you pitch that thing and you realize that some program manager did it back in 87 and you're all bummed because you're like, oh man, you can't come up with an original idea and these four walls that somebody hasn't done it previously that.
[00:00:52] Then then you just after you get over, you know, the being hurt that you know that your idea is already been done. Then you go talk to some of the original contractors you go talk to some of the sita's you talk to the folks that were there and figure out what is different and then and that is part of the catechism that is part of the what is different.
[00:01:13] Now that will enable this to work in a way that it didn't work previously.
[00:01:18]Best ways to Enable Robotics [00:01:18]
[00:01:18] Ben: [00:01:18] Yeah. The catechism is I think a. Our full set of questions that people don't ask enough outside of DARPA and I'll definitely put a link to it in the show notes. So I do know we're coming up on time. So as a final question, I want to ask [00:52:00] you've been involved in robotics in one way or another for quite some time in Academia and in governments and start.
[00:01:42] And it's a notoriously tricky fueled in terms of the amount of hype and excitement and possibility versus the reality of robots coming into especially the the unstructured real world that we live in and why do you think of there? There's a better way to do it from sort of all the different systems that you've been a part of like is there an entirely different system.
[00:02:10] What would you change to make to make some like more of that happen?
[00:02:16] Mark: [00:02:16] I this and I hate to say like this. I don't know that there's I don't know that there's much I would change. I think that right now especially working in robotics. That is I look at the. The capabilities the [00:53:00] sensors the all of the enabling work that we have right now in terms of machine learning and autonomy and everything else like it.
[00:02:41] This is a great day to be alive and working in the field of Robotics in a way that you know, and I'll feel like the old man is I say this but you know, I started this back in the late 90s early 2000s and frankly when I think of the tools and the platforms and sensors that we had to work with. Um that that you spent especially my experience was a grad school grad student experience.
[00:03:08] But when I remember how much time we would spend just just screwing around with sensors that didn't work right in platforms that weren't precise in their movements and you know, just all the other aspects that make robotics robotics and I now look at today the fact that you know, we've got kind of.
[00:03:30] Bischoff platforms that we can go find that [00:54:00] you can use that that for these lower low-cost platforms. You can really dig deep into research areas that are still just wide open. The fact that now, you know in the mid-2000s if you wanted to do a Thomas car research you needed to be especially or basically.
[00:04:03] They don't know how to work high power crazy high power servos and other things like that. Now you go buy a Prius like or Tesla or something, you know what I mean and you're off and the platforms built for you. We you know, the the lidar the computing power and everything else we have today. I might answer your question right now.
[00:04:23] I don't know that I would change a thing. I maybe naively believe that we have all of the tools that we need to really really. Make dramatic impacts [00:55:00] and I believe we are making dramatic impacts in the world that we're living by enabling Automation and autonomy to do really really incredible things.
[00:04:43] The biggest thing is is for folks to to go back and to kind of along the line of the last line of questioning the you would have had as far as you know, forgetting and remembering the things that we've done in the past. I find that some of the best ideas that I'm seeing that are coming forward into Robotics and autonomy are.
[00:05:01] Ideas that were really born back in the 90s. We just didn't have the computing power or the sensors to pull it off and now we do and so it's almost a go look back and look at you know, kind of create a Renaissance of us going back and looking at some of the really really great ideas. That just didn't have their day.
[00:05:23] Back when you know when things were a little more scarce in terms of computing and algorithmic complexity and other things like that that we can now address in a really kind of powerful way that [00:56:00] is quite a note of optimism. I really appreciate it mark, thank you so much for doing this. I want to let you get on with your day.
[00:00:06] I've learned a ton and I hope other folks have as well. Absolutely. Well, thank you for having me on I appreciate it.