Transcription provided by Huntsville AI Transcribe
So what we’re talking about tonight, the government has a program called SBIR or SBTR, Small Business Innovation Research. I think everyone has SBTR. Okay.
Technology transfer. We’ll go with that. It’s on one of the documents that we’ll look at later. So, right for tonight, what we’re focused on, there are several government agencies overall that do this kind of thing. The ones we’re looking at here are more of the DOD Department of Defense type things.
You can think of them.
They’ve got three phases to these.
Phase one is think of that almost as a seed fund, small piece of money, small, 100 to 140 thousand dollars. Usually it can go up from there. It can get a, what do they call the bridge part option?
This is an option, I think.
So if you do well enough, usually the phase one is about a six-month effort on one of these topics that you might propose on. And if you get funded, you do that, you do a good enough job.
They’ll say, hey, we think we want to continue this to phase two. And then that’s where they kind of exercise whatever that option is.
Give you like a bridge fund to make it to the phase two.
And at that point, you actually put in another proposal for the phase two. Phase two is generally a growth forward of magnitude.
It’s about two million dollars in two years.
So that’s more funny, but that’s why they get you to enter into your bills. Yeah, I can’t do something. I can’t do anything. Yeah, so phase two is also an important knowledge study.
And then you have to have at least a prototype in the phase two.
Ideally more.
Right.
So that’s where they want real stuff for phase two.
They’re paying you real money now for real stuff. Now that they qualify, but she kind of knows some things. Some of the phase ones are somewhat interesting.
Occasionally they want brand new ideas of things that might be, might be impossible. You know, I mean, sometimes they really want to push the envelope. Sometimes it looks like, you know, they’re just trying to find a way to take a new technology and integrate it into their, you know, something that they already have working. And over the last few years, it’s gotten much, much better as far as the quality of these topics, you know, as far as what they’re putting out there. About five years ago, we did our first survey of these from a console AI perspective. And there might have been 10 topics.
Half of them were probably things that were provably impossible. Maybe two that if they had Googled were just something you could go by today. You know, about a new development thing, they just didn’t know it, you know.
And maybe two a year that were like really interesting doable things to advance technology.
What we’re at now, across the, we’re actually on the third round that get released a year.
And as in usual, what happens is you get a lot of topics the first two rounds and then round three is usually a little thin. And it’s a little thin again this year.
So total number of topics for small business side is 38.
Total number of topics for the, actually I think I got both of these on the other area. I mean, moving this tab over so I don’t get lost again.
And I had eight topics for the technology transfer.
The other caveat is that for the small business part, to be able to put in a proposal for this, you actually have to be a qualified small business by the government’s version of what a small business means.
I believe right now it’s, I think it’s less than 500 employees, which most of us would think is a pretty big company. But when you compare that with other like big DOD companies, you know, like your Lockheeds and your Northropes and whatnot. So yeah, yeah, it’s 500 or small. So there’s that. The other side of the coin, these STTR topics, for those you have to either be a, like a research-oriented institution, like a nonprofit or a, or a, some of these government labs, some of the labs, some of the other things like that.
Some of your colleges have, what was it called at AM? The, uh, miners.
Right. So that’s it.
And search.
You want to form, there’s one that’s, one photo there, one photo there. And there’s something similar at UAH. I can’t remember what they call there.
It’s on the back of the ITC.
And then they also have there was research institute. Yeah. And then we’ve worked with, we didn’t get it off the ground, but over at the creation force, we were close to being able to pull something over that actually had Carnegie Mellon.
I mean, most of your big research institutions have an arm of it specifically to partners with industry to try and get, because they’ve also got students that need to learn stuff and also wouldn’t mind making a little money on-site. And the interesting thing with STTRs is generally some of the research institutions you might work with have different opinions of who owns the data under who owns the rights to the thing you do.
So, up until- Well, there’s universities who say everything that the professor and best say on.
Yes.
There’s a university that says everything, a partner company, that’s as far as STTRs.
They all, yeah.
And then there’s a university which is quite famous that says anything you invent will help you start a company so you kind of be successful. Right. You know, the university, that looks like it for your standard.
Yeah, because they’re actually the people who came up with the stamp of lean, entrepreneurial program which ended up becoming ICO. Okay. And the whole purpose is I want you to be extremely successful. Right. So the GK- Because if you are, you’ll donate money, right? So then they turn it all completely around. I think it was said we’re going to be greedy, we’re going to be, oh, we’re going to help you. Any good idea will help you build this successful. And then, and then, uh, basically that’s Silicon Valley right there. Yeah. Oh, this is it.
Yeah.
If you get greedy, you might get destroyed under the stick. Yes. So, uh, this set of topics dropped today, which is very adventurous, put together a talk on topics that you haven’t seen yet.
I gotta tell you that’s, uh, yeah, I’m trying to get on the Hudson Alpha network so I can actually help you. On the refrigerator door. Yeah. Yeah, me too. I think it says 601 genome. That’s the address.
Yep.
Oh, anyway, if you ever come here and get a lot of password, do you have to have secret quarters?
It’s Virginia. Virginia is not a password. Um, generally your phase one, uh, is, is unclassified. So the first part is usually unclassified.
Most of the topics, but we’ll tell you that hey, for phase two, we expect this to be done in a classified environment with classified data.
There are organizations associated with whatever, was it SBA?
I can’t think of the name of it over Archie government organization that runs these. Um, they actually have programs that can mentor you as far as how to write top, how to write proposals, how to put together your submission. They’ve also got places that you can go to figure out what do I need to do to get access to, you know, how to, if I don’t work for a company that is holding or has contracts associated with it, uh, is going from a non-secret clearance to a secret clearance is not an overnight thing. Well, how do you, you know, so there are places that help with that as well. So what I was going to do at first, uh, let me get a little bit closer.
Okay. Okay.
Some of these I can try to expand the websites, uh, could seriously use some help, from a, you know, experience given the number of scroll bars and we haven’t even hit the full number of scroll bars yet. So usually, uh, what I’ll do is go in here and I’ll drop the word learning.
I also can only head to one third of a previous person, which has your PM over there.
Can you do this backwards?
Can you find all of the proposals that don’t have anything to do with any, uh, machine learning?
So you want to put crime after that?
And I’m sorry. So let’s, let’s take the groups of people that are looking at this tonight.
How many of them want to do a, uh, every budget?
What’s the one solution? And nobody’s going to show up to my meetups. So, uh, they could not figure out whether to call things artificial intelligence or machine learning.
There’s also other things that have the word artificial in them that will trigger.
There’s other things that were intelligence that will trigger. So intelligence has a whole department for different branches. So learning how to begin the best work. Uh, there’s also, I think we’re gonna, we will also look at, they have, uh, other ideas for modernization priorities across DOD as a whole. So overall, they really care about 5G advanced computing and software, you know, and they’ve got a lot of categories. Yeah, these are all the, uh, yeah.
The one we’re looking for is trusted AI at the top.
I wouldn’t write to the top two.
Um, there wasn’t any, I know.
Was there?
Yeah. Oh, yeah.
There you go.
Oh, huh.
Yeah, and uh, so I can’t even figure out what, what category or two.
I just did a search for artificial intelligence and, uh, the query revolved and gave me about seven different solicitations and that for grants. What’s a WFOP?
Uh, wide-angle field of view.
Wide angle.
Oh, yes.
All right.
I know. This is an interesting one that you, you know, I’m interested in.
Let’s just start from the top. I haven’t seen, well, the thing, uh, Jonathan mentioned at first, uh, you will see on these, there’s an open date and a closed date.
And you’ll notice that, uh, open is not yet.
So generally they will drop these in a pre-release state, uh, and they will typically stay there for a month. In this case, it’s not quite a month. That’s through me at first. So they put these out pre-release today. Then you have up to, uh, September the 18th to attempt, I mentioned a news word attempt on purpose, to contact the tech, the topic point of contact, uh, with any questions you might have about, you know, what, what’s written or whether a particular idea or avenue of research would be advised or disallowed or anything like that.
Uh, you do have a much, much, much higher chance of being awarded a, one of these.
If you do actually have some kind of a, uh, conversation or build some kind of a relationship with the doctor, Jonathan. Um, yes. Some that we’ve seen, Air Force, I believe, is it Air Force or Navy? I think it’s, I don’t want to say Air Force. I don’t think it’s Navy. Isn’t it?
If you don’t talk to them, you don’t have a chance. Well, there’s one that’s, if you don’t talk to them, have a chance. There’s another one that won’t even let you talk to the topic author directly.
You have to submit all of your questions, which they get posted with everyone else’s questions, which gives the rest of the world insight into what you may be thinking.
So, you have to be careful about what you said formally. If you have the genius idea, you don’t want to ask them if that’s a good idea. Right. Because everybody else would go, that is a great idea. Let me put that in my proposal. It’s now public knowledge. So, a lot of times these, uh, not sure. See, so if you have a foreign malign influence, okay, you have to, you have to be careful about how you do that. It’s a joke there. Right. One, if you post it on the web with the government, then you’re, anyway, I’ll stop. So, we’ve got, anyway, so that’s interesting. Oh, you were catching this, this word down here, foreign malign influence. I got it. So, the first one, we’ll cover these pretty quick unless we see something that everybody’s like, oh, I really want to jump into that. And of course, be aware that some of us here may actually be working for companies that write proposals for D6. So, I am not going to tell you what I might be working for in the future.
And I expect y’all to also not tell me if you happen to be working, if I don’t want to fail, I’ll let that work. So, the first one’s actually, this is the interesting part. I didn’t bring this up. There are some topics that they will put out there as an open topic that is basically bring your own, bring your own ideas. We want something in this kind of area and you’ve got some funds set aside, but we’re not smart enough to know exactly what to ask for. So, they will put out things for open topics, looking for specific, you know, kinds of areas.
Sometimes they’ll give you like a shopping list almost. We’re interested in these eight things.
You know, I may have radar and it does discrimination.
I may need to worry about power consumption for optimization, something or other.
I may need to work it. There’s all kinds of things.
Now, we’re pretty sure AI does these things.
So, you know, and a fun thing there is we can hit one item on the shopping list much better.
What’s PBI?
I don’t know.
In my head, I’ve got potentially adverse information about those. That’s correct. Given that especially the key works here.
Well, I like the three words in the middle.
Training, synthetic, environments and serious gaming.
Okay. So, this is basically talking about campaigns brought out in with an election coming out. This might be interesting.
As far as being able to generate significant amounts of information that appears to be written by humans and written by other organizations.
Yeah, go back to the title again where it talks about the disinformation.
Are they talking about how to do it or how to protect? Don’t know yet. I think they’re getting there. I think when you say foreign malign influence, you have to add because disinformation is provided. Right. So, there are campaigns that target people with disinformation and brought to the end of specials or mitigates.
Interesting.
So, games and synthetic environments bring an immersive experience that improves learning and user experience which they apparently apply here instead of your website. So, so, is anybody else here had to take training about how to not click on links? And all of that.
So, I think what they’re looking for is some kind of an overarching kind of like that simulated game that you go play.
I’m going to email the other one.
And it said, I was able to double for a hundred dollar prize to finance the following survey. And the last three digits of the email address was.ch.
That’s fun.
First thing to check. Right.
So, this one, it looks like phase one is basically a almost like a domain survey of for this kind of training, for this kind of training, what is available out there currently?
What is the state of the art? And then, you know, basically put a proof of concept as far as how you would build some kind of a training scenario game kind of a thing.
And then phase two, let’s involve it and build a prototype.
So, in this one, in this case, you don’t even have to have a prototype at the end of phase one. You have to have a proof of concept.
So, it’s pretty interesting.
I could see the other part of the SBIR type approach.
They also want you to have some kind of a commercialization piece in your proposal.
They plan on giving you some seed funding to go build a thing. They don’t want to keep giving you money every year to support it in advance. You know, they’d like to help get something off the ground. And then kind of like what Dr. Borty was saying about the Stanford approach, where let’s get you going and then we get free use of this thing, but we all have to pay to support it after it’s already filled it somewhere.
I’m always asked the question, if you did this, who were the customers?
And that’s the hard part.
For this one, there’s one later that I think is was really hard for me to put a finger on where the commercial part would be.
In this one, if you made it some kind of a modular piece that you could migrate from one domain, in this case, it’s foreign military or whatever the propaganda kind of thing, if you picked that up and put it somewhere else, maybe a different domain for your immersive training.
See, I think it’s going to be a constant effort to keep this thing up to date. Yeah. Once they figure out what you’re doing, they’re going to have to find a way to get around that and do something different. It’s going to be a constant struggle, offense and defense. So that one, I’m not quite as interested in. This one is foreign.
Oh, hey, there was one word to this narrative. I’m going to have to show it later. Pedagic geological. Pedagic geological. Preceded by Pedagic. I’m not sure what that means.
It’s teaching methods. Professor, actually turned down the job for that. Because I got much better off of it. I was. I was going to make it the other day, and the guy actually used himself as the junior doctor. I realized he was at a table and I had been rotating. So anyway, he’ll always be the junior doctor for us. In this case, for this one, characterization, I think of R2 require targets using advanced machine learning methods on wide field of view staring data.
Space platforms should give you a hint of what we’re talking about.
So if you happen to be on a satellite somewhere, you know, actually, no, this is actually different.
I remember this.
Okay.
This is the problem you want.
Very good.
That’s what we’re going to have to do. Right. So, currently, there’s a command with the Air Force called space.
I think it’s maybe Space Force by now. Who’s job it is is to keep track of all the crap in space. No, there’s a lot of it. And we keep pushing stuff out there every year. And some of these lower Earth orbit pieces, it’s a lot of pieces. You know, you’ve got a lot, which I think I just said that, but anyway, what’s the first?
Of course, yes, they’re careful to mention that SpaceX is diligent and self reporting, but they’re putting out the trajectories and all that kind of stuff, but I’m assuming they might make mistakes. You also have foreign entities like China that might be potentially uncooperative and also pushing out 12,000 plus satellite conom’s constellations. So, and this is where you got to go push the magic button. You want to do the magic button?
I’ll see if I can figure it out.
It’s the light switch that doesn’t, on the right side of the door. SpaceX is already got it.
But there you go.
SpaceX has already got, I don’t know if that’s awesome up there. Well, that’s what it’s saying is they kind of where they are really going.
Imagine you had a part company doing that.
That may not be as nice. I’m just going to say that in my recent work history, I’ve worked on dual sensor assignment problems. And I understand the algorithms for making them both efficient and accurate. And when I say efficient, I mean orders of magnitude faster.
Okay.
I’ll leave it at that. Somebody wants to work on this. I can help. See, now the point is, is if you have six stereo cameras, and they’re all saying approximately the same field of view, you now have, I’m going to put five things up. You’ve got an image with these five things, and you’ve got another image of these, of the same five things. Okay. Now you have to correlate which one is which.
Okay.
Yeah.
This one, this is, this statement is problematic. Oh, yeah. Occasionally, you come across topics that appear to be written directly towards something that somebody is already working on building for some company that would have to have its own data.
Also, a signed letter of support, which when we write these, we have to try to find letters of support from other, other users or companies that we work with and say, yes, if this was funded and built, we would look at, we would be interested in integrating it in our product.
Yeah. They’ve got to be using artificial intelligence to acquire and assimilate all that data behind people who wrote them. That’s long enough. They’re already in the industry, they already have a data set. Right. That’s kind of the infiltration.
That’s the data set.
Right.
So if you go figure out who is current, what small company is currently working while in a film, when you stare at it and blah, blah, blah, then you could probably figure out who is going to get awarded the top.
Pretty quick.
But the other thing I’ll bring up, occasionally they actually award more than one in Phase 1 award.
Yes.
That’s something that I’m trying to come to this point. Yeah. A lot of times, again, the interesting idea is that they didn’t expect. So even if there is one that is geared towards, it feels like it’s geared towards, don’t worry about putting your own thinking in. You may have an idea of whoever it’s not for and haven’t pitched yet. And you may actually. See there, it says edge devices.
The bottom sentence, the very bottom sentence. Down here. Yes.
Yeah.
Well, put your little bit to the right. I’ll see an edge device.
Edge devices. Right.
Okay.
Makes they don’t want to run off to some high-performance computing.
You want an edge device you can put in space. Okay.
To do the computation. You get a clear picture up there. Well, I’m just saying.
Well, you got a, you got a planet in the way from the other side.
Well, you know, you could say.
It’s a multi-phase process.
If the end result is an edge device, but in the beginning you can do whatever you have to do to get the algorithms and things to work.
Eventually you’re going to have to translate this into an FPGA or a collection of a Raspberry Pi or something like that that would go into space.
With very low power requirements and rad hard.
And here’s where you get three different scrollbars.
I remember it was a better one computer.
Oh my gosh.
So those are the two right now that are SPIR and they’ve got released.
If you other thing we flip over and look at the STTRs.
Which actually was this.pore. Pretty friend soft tissue. I’ll try it. Yeah, this was the dot three.
And then the dot. So the STTRs, they can’t use numbers.
So they have to use letters. And I’m still a post-mathic college.
Yeah, now there’s got one, two, three and these go ABC.
And for this one, it can be the same.
Filtered.
Well, I just saw a fun one.
It had to do with physics and radiation and applications relating to AI. So that got my attention. Was it the radiation assist? I already looked at one of these. Yeah, there’s this.
Is this what you were talking about?
That was one of them.
I had a few others like that.
That looked like one had to do with plume assessment from like say a mushroom cloud. I’m sitting here going, hey, bigger or less, smaller.
That looks very doable.
Yeah. So again, this one would be you’d have to tie in with a research institute or something like that. Or be a research institute. Sometimes you can be both. Yeah, I know. I don’t know what you’d be prime if you’re the research institute, even though 501 seems to be.
I know audio, I’ve shook that tree really hard to try and get leadership to see it our way. They don’t have them. So this, this is kind of interesting.
I don’t know enough about radiation physics.
But I think what they’re looking for is something that can run like so much of the age device that Phil was talking about a second ago, lightweight, carry it with you. If you were in like, in my case, as a first responder, if I went somewhere, there’s all kinds of reports coming in from different places that have all this mumbo jumbo and technical.
Well, just laughing that they, they say, can they, they ask for the biggest.
They, they take care of it. Yeah, I got that. That got me. Really?
Yeah. Yeah. No, it was usually this is, yeah, expecting a little more formal, though.
Oh no.
I don’t know.
They have fun sometimes when they say it. So the other fun part about this, sometimes these, these topics come from things that they need as in, you know, they really, really need this. Other times they come in as in, hey, I am head of the department of X and we’ve only got some topical truth. Oh, partner B has 28 talks.
We had to go create some topic truths. You call that an ego problem? Yeah.
What?
Oh, okay.
Yes, of course, we’re on the receiving side of this, you don’t get to see all the hypoesthings, all that kind of fun stuff.
So this radiation physics and thing. I’m going to interrupt and tell the short story of this from the heck of it because I know the story. When they dump brown stereo over here, they have operating reactors. They have a two that goes from the control room down to the reactor. It’s one piece of pipe. They have control wires that goes down to the reactor. The electrician or the technician is stuffing, padding, okay, in the pipe. All right. So that the air from the reactor doesn’t come up to the control room, right? Radiation. Okay. How does he tell us to see if there’s an air leak from the reactor?
Does anybody know what air leak he says? An air leak.
There’s a, how do you know that there isn’t an air current coming from the reactor coming out of the pipe where he’s stuffed with this cotton around the wires to stop the air leak? I don’t know. There isn’t one.
Don’t you try to go up something really small like helium?
Mm-hmm.
Don’t you go up something really small like helium?
Because that can get through when you’re pre-controlling. You’re about to imagine holding up the carbon, the fiber area. A body plane.
Then you see if it’s a plane plane.
Let’s see.
Okay.
Okay.
All right. You can also wet your finger. I get a grasp of it. He’s down holding up these holes and I’m holding them. Then he sets the cotton with fire. Okay. It starts burning up the control wires and it burns for about six hours. And they don’t know how to put it out. They call the act inspired department.
The guys come out and they go in and they said there’s a fire door. But we don’t know how to get it out. He says, the boys have been in the holes and pulled water on it. No. This is a true story. I’m not making this. Oh, I believe me. I’m just cringing really hard.
Okay. They tell them there’s bare water or something there. Yeah. They shut the reactor down. It was down for about three years. We thought they got everything back together and got it off and off. Then they get a different electrician. I bet they did. Go look it up. It’s a brown-stairy fire.
It’s a true story.
It’s an interview you’re 25 years ago. Do you want to let it downstream of Brownstair right now? Just insert an inflatable, get it, push the roll of insulation, inflate it. Put a cap over the outside and it’ll die. Okay. You pick up a thing of plaster and you pour it in there.
Okay. You can put a little bit of death in the pipe in the tube and you fix up a thing of plaster and you go push and it fills it up.
There you go. There you go.
The brown-stair.
There’s things to do.
There’s different plants. Okay. If you want to be super secure you can put two depths in.
19-7.
Let me bring us back.
So it looks like what they’re trying to figure out to do in general, put down those sentences, how do I build a model to take a really complicated journey for radiation and explain it to a 5th or 5th grader, I guess.
There you go.
This is something ChatGPP does like eight million times a day. I can’t tell you how many times I’ve asked ChatGPP to load something to explain this to me like 150 degrees.
Yeah, it’s really good. Apparently it does a lot of 5th graders. I know because you need to learn density functional theory while I’m like, you know, what’s the point of field theory for that?
He’s like, oh no, ChatGPP teacher.
So again, this is a proof of concept like a paper kind of thing for phase one and phase two is when we actually build a prototype along with this is the fun part.
Oh yes.
So this is a thing that’s supposed to help you learn about radiology things in your area without how do you test that?
Well, either detect it or you send it to a browser.
Yes.
Either you need a detector or you need some Monte Carlo simulations to model like from, okay, these are all these scenarios.
So you need a detector or a network of detectors and then sensor fusions another way about it.
But I don’t know.
So what you do is you write one proposal for this one and one proposal for the other topic we just saw for a synthetic environment for simulating a game thing. And you use that to test the people that are doing the thing here.
And you wind up with at least maybe one thing if I could use some of it.
Would you figure that out?
Oh boy, he’d probably tell them.
They probably think they probably love to have the two things kind of tied together. It’ll make the agencies look like they’re working together. Yeah, they love that. I know I’ve asked them. I remember asking DLE, hey, would you ever want to work on that? I said, oh yeah, we would.
So then there are other topics. These have been out for a while. Some of these close fairly quick so you can really do less.
So there are three rounds that usually get dropped.
And then sometimes they’ll just drop something that it’s likely came up with some stuff inside if we don’t want to wait.
What’s an A-lane fuel? I was wondering now. I’m going to look at that.
Don’t know.
So that’s why it looks like it happened in 1960.
Is older fuel some sort?
Let me get filters. There were some very interesting ones in this set.
So that’s why I wanted to… I made an aluminum and hydrogen chemically bonded together. It is a fuel that’s comprised of aluminum and hydrogen which I think chemically bonded together. It’s a solid. More like aluminum. Long on the solid, it’s like a solid rocket fuel. It’s like a fuel solid sort. So the other thing I’ll bring up and see some of these were open actually. It’s six.
It looks like there’s some special forces there for the army. What I’ve learned is if you see X in front of the topic title, it’s usually… Usually I’ll have to play. No, usually these instead of writing like a proposal and going into it, they want you to come pitch. And so we just have to start with a pitch competition thing. Yeah, they have those. So if you may be not great at writing but good at pitching and powerpoint, that might be a good way to do it. The other thing is generally for these, they’ve got somebody with a checkbook there. So right now you write a topic you wait, I don’t know how many months before you find out if you want or not for these. The turnaround is much, much quicker. Anybody want to do one of these? What’s five minutes of my time worth? Oh, you’re talking about the $2 million. $1.7 million for the missile defense agency over six years. I remember this. What’s 15 minutes of my time worth?
Depends on the… $140 million worth of running the computer center at NASA Goddard for five years. The most I’ve ever made in 15 minutes. Okay, so if you want to… I’m happy to take the presentation.
So… Well, what?
This is one of those that are… Have a wide open.
They got some topics that they’re interested in. Denied, I don’t even know what this means.
I know what a swarm tracker might mean.
So if somebody had put a bunch of drones together and had them kind of as a swarm, how do you… It’s hard enough to track a single, much less multiples that may or may not be working together.
I’ve seen some… I’ve seen this in the past.
They’ve seen one tracker.
Yeah, they may have not got some good… Sometimes you see the same topic might get recycled a couple of times, if they don’t get good… Yeah. Let’s see. Can I share your network?
Streaming my trip magnetics.
Okay, so… There was an article today that the Russians are using a tube under the drone and it has a fiber optic cable in it that must be 10, 15 miles long. And they’re flying a wire to the drone. And they’re flying a wired drone. So there is no radio control to block. No. I can tell.
Perfect idea.
It’s worth the more… Not a big deal. We’re going to close this one down. Jump to one of the covers. Next one up. And we’ll hit some of these quick, because we’ve got 10 minutes left, dynamic generative large magnet bottles for continuous situational areas.
And it’s got hallucinations in it, so they know that’s a problem.
One thing I will mention that I don’t think we covered yet. Sometimes you come across these topics that will say these are direct to phase two. Phase two.
Yes.
Generally what they’re looking for there is somebody that has already had a prototype level or some adjacent product that they might be able to modify or something. So they’re not… They don’t want to start a distraction. But you also start a phase two level of funding. So, you know, this was actually fairly acceptable for 18 months.
Two big. So they’re actually wanting a little something a bit quicker.
How much do they review?
Do they review what? If you’re a T1 project care, how often do they review it?
How do they review it? You want to talk about that, Phil? I want to. You want to talk about that, Phil? I want to. I want to. The one that I was involved with, we wound up… After you get awarded, you’re basically having contracts that go together to do work. And so you have certain contract deliverables that you have to give them like what’s the money, you know, based on the contract, as well as what information they want, what kind of reports they want, you know, the financial reports. Okay, so it’s variable. It isn’t normally it’s monthly, but that’s generally what you see.
The other thing you’ll find if you win one of these is that your point of contact on the other side might actually be responsible for four or five other research development items at the same time.
So along with maybe a full-time job, I’ve seen that once where we had a hard time getting a hold of the person that was responsible for our contract.
So this one is the directive phase two. They’re trying to figure out how to mitigate bias and model outputs, making sure… Of course, we were just having a discussion on this when we first got in here about fair and unbiased information.
And the question was, well, reporting to who’s judgment is fair and unbiased are we not? So anyway, this got all kinds of rabbit trails you can jump or follow.
But this is pretty straightforward as far as there’s most likely things that do this already.
They may not know about it or may not know the right words, put it to Google, find it, order from what not. That’s fairly interesting. Another thing you have to also… Okay, science directs and then one of the UI.
Looking at what references they drop in VGO4 as well to let you know what’s going on.
How much they actually investigated this themselves. Also the types of references. We’ve come across that before where we’ve got one topic where most of their references are more trade material brochures, things of that level.
And then we come across other topics where their references are basically scientific, PhD level pieces, paper. Oh yes. You write those proposals very differently.
If there’s somebody that’s normally cruising through thesis papers, I’m not just going to blow something over with a five chart quick thing.
Likewise, opposite land.
If I give somebody something with so much technical terms in it, and what they’ve been reading is a very extremely high level PowerPoint deck that they got from a product document somewhere, it’s not going to go well.
The audience. Yeah. And then this other one. Wait, let’s finish that one.
Okay. There were other things here for… Deering, face to… Interesting. Direct to face to, that’s what they’re calling DB2.
And then for face to, they’re basically saying do that.
Face three, a lot of times is kind of where they expect you to deploy this after you finish face to. It’s kind of hard to get into a face three. There’s a lot of times you are asking like a big company that has a big contract, getting your thing added to their product. Good news is they generally get a lot of good credit with their government customer for actually helping do something. One of the difficulties with these proposals is this totally use. If you come up with a solution that’s great for the government, as no dual use, you’re in a bad shape.
And I don’t want the government doesn’t want to fund the development of some great big thing, then it might be have a commercial application. The fun thing I’ve just noticed, or I’m probably not the first person to come up with this, but they have to put a commercialization plan in your proposal. Yes. Here’s some good hints for what’s put in your commercialization plan, because they’ve already thought through where else could this be used.
Maybe. I think it’s very about to get killed by the scroll bar thing.
Dr. Bordelger showed me at one point that NIH has the best approach, the most robust approach to a commercialization plan.
Is there a guide manual?
It’s just tight. That’s NIH?
I think that’s NIH. She showed me it.
It’s been a few years, but yeah, I still have it on my computer. This is pretty interesting. I’ve come across this before several branches of the military actually had a formal process for making decisions.
Some of them include like high level trade studies, others have drilled all the way down to looking at information, timing this impact of the wrong decision, things like that, and then drilled down. So the Ukrainians just did this, right? Read the last to create decisive links to strategic surprise. Yeah, they sure did. Yeah, the last week before I’m saying, right? The Russians took down their defense for a better thing and they agreed it’s probably a tax. Thank you, sir. That’s interesting.
This one again is a direct case to Alderley.
I just took an AI machine learning workshop from NVIDIA at Wichita State for two days and we came up with these 94% and 96% success rate matching and things and they totally ignored the error. In other words, if it was 96% fit for the model, they did not do anything you explain the 6% difficulties. And if you were trying to build a useful tool, the first thing you want to know is what did it not get right?
You want to know explicitly as much as you can about what didn’t get done right. Yeah, but you could never get to it. Well, that’s the point.
So the things that didn’t fit, you want to understand what they are. Okay. Because that might be, if everybody else is doing the 94% and you see something in the 6% that you can exploit, they’re not looking at that, but you are.
Yeah. Okay.
They got an answer to this thing already. They’re already dating her. So, it’s pretty interesting. Which is good.
Yeah. Open topic, skip that.
The first engineering, open topic.
So this one I’m slightly familiar with from some other stuff where you see something, you got to figure out half work.
So, with limited knowledge or limited visibility, enter them in the things.
Chinese and Russian. Oh, yeah.
Not only did they reverse engineering, they could move you another next week. Who was I talking to? Who was?
Still trying to figure out what exactly they want. Oh, never mind. This is, that’s why they don’t take that quote. They want it to know the topic with some different thoughts. So, the first one they’re looking for, and this was something I think they got a system that actually does this. I don’t know if it’s a Marshall where it’s like a big line on our standard thing.
Oh, they can scan something besides a little isopag.
I mean, it’s a big thing.
So, if you need an exact drawing, you know, the big large piece of equipment rips around it as you would.
It’s, this is kind of taking in another level to scan it and then figure out what components it’s broken down into, what not.
So, they’re looking at that, uh, training platforms, simulation requirements, design, figuring out how to do design optimization on a bad quality.
They’re looking at some for some, for the lead, letter, letter of county, armory, deep ball. Oh, this is a specific site solution.
Yep. Address, science, use, plugins.
Interesting. Okay.
It does give you some, I just, I don’t know if I, I don’t know whether I should pronounce that lead or lead. Pretty sure they were thinking lead, like, yeah. Uh, it looks like they’re looking for something to partner with. All right.
Next one, I think we’re close to the end of this.
These are also co-contactments. But what we don’t, but the other thing I was going to show, before we run totally out of time, but we’re over a little bit. Uh, on this site where we were, if you hit solicitation info, those solicitation documents, instructions, you can actually download, like what I had open is the 24.3.
It’s a PDF file.
It’s got all of the topic info that we just looked at.
So, I’ve got that here.
I actually going to download that somewhere.
Scoped out. There. And then I’m going to hop over. I feel like, no, I don’t want to search. I thought you were too much at CTT account. Uh, which right now I’m logged into our collision force instance, which right now I’m just using this play area. And if I go find my downloads, I’m going to take this whole thing and shove it up there.
And this was like a 266 page big document of stuff. Uh, so let’s go with, uh, anybody online is better at this than me.
Feel free to throw ideas out by making this up on the fly. What I’m going to do first is, uh, using the attached document. Uh, this topics. Blooded that reference.
AI slash. Yeah, let’s see what happens.
Says it’s reading it.
See if I can actually.
You just showed me dangerous. So it looks like. So it looks like, uh, also I didn’t realize DTRA is defense throughout production.
Hey, I was not worried that. Uh, it’s hard to acquire targets that come from that staring the, the, and whatever.
I can’t say where I filled you.
Uh, let’s see.
Can I take control systems training games?
That would be the one that was the synthetic environment. And then the other piece, you know, for the, uh, alignment as well as refer, they all reference the same doctor because I’m the one.
Uh, what types of AI would you?
Suggest.
Solve.
Problems.
Okay.
So one of these was an OPE, OCR, maybe an indirect commission. This one they think would be convolutional because it’s basically in this space kind of stuff. Long as reinforcement learning training. Graph neural nets for the space domain awareness.
I mean, you get where I’m going at here.
I’m pretty sure I could upload a couple of, uh, submissions that I haven’t provided before and say, could you build me a submission to topic X using this perceived with a blah, blah, blah. Um, I can’t remember how many years. When was, uh, I think it was GPT two that had come out and we actually did a session here, where we trained it on the 15,000 SBIR topics that they didn’t put out there previously. And you would give a sentence and it would generate a full topic description. Phase one, phase two, you know, that was a couple years ago. So that was fun. It would come up with some crazy stuff. Some of it was real. If you could do that then, take the insurance and just go out and find who could do that. They don’t have to think it up. All you can do is turn it out, put the blocks together. You can take this stuff further. Say using this exercise. Do that.
I mean, it’s, uh, I mean, you’re basically using the chat to bootstrap itself with additional info. Which is usually the first thing I do when I hit a chat GPT company, if I don’t know how to ask the question, I ask it how I should ask the question.
And it’ll usually give me some kind of format that’s probably worth it. So that let’s shut down at least, see, we still got everybody closely utilized.
I’m going to stop the recording and then we can hop into any kind of questions, answers, things like that. I’m not saying I have any answers.