 Talk about the future of robotics manipulation with open source. So please go ahead, Tyler. Sorry, sir. Yeah. Thank you. Um, so we are the, um, the people behind the library move it, but more generally, I'm going to talk about a little bit of the history of robotic manipulation and need you into kind of like where we are now and and what, um, what we're doing in this field. So the big deal with robotic arms right now is logistics. Um, these are just some of the companies doing things. It's a lot of pick and place in sorting facilities and warehouses and stuff like that. It's kind of the new hotness manipulation. Um, but a real basic question when you think about like, where is the robotic manipulation? In most cases, the best robotic companies are ones that are getting rid of the robot, like the robotic aren't robotic arms are really hard. And various companies are finding mechanical solutions to do similar things. So here's an example, um, flippy and creator. They're both companies that are creating robots for Uh, making hamburgers and, um, one is building an entire mechanical system creator for the whole process. And the other than flippy is they're, um, they're, they're building robotic arms simply just to flip. Um, another example, um, is making coffee. Many of you probably seen robots resemble the one on the right. The one on the left is a company from the cafe X. It's coming out of the barrier area. I think that is trying to use a robotic arm for making coffee. But what was, what's, what's the real dream of, of, uh, robotic manipulation and that is a multi-purpose robot. It's not a does one thing like the coffee robot or the hamburger flipping robot. It's, it's really, um, it's really a general purpose robot that can do all sorts of things. So background on robotic manipulation. Um, there's sort of the history of what's been important in different time periods. There's, uh, example of the, the Puma industrial arms in the 60s and then teach pendants were a big deal. Um, teach pendants are still a big deal. That's how you train is the term they often use to you, you put the robot in a state where you can manipulate it by hand and develop some canned motion that you can repeat over and over to do some tasks. Um, 80s machine vision started to begin the thing and 90s PC based control. Um, and 2011 ish now ish co-bots are a big deal. Co-bots really just robots that operate in a collaborative space with humans. Um, so where have robot arms been really successful. The past few decades, they've been really successful in large scale manufacturing. Um, in those cases, you're often just doing the same thing over and over in a really controlled environment. So where, what, what, what is next? We see the future as being mobile manipulation. They're sort of the next frontier. Um, that is some form of manipulation with some form of mobile base. Um, here are three example robots. Um, some of you have probably seen. So robotic arms, robotic arms from our perspective is a, is a real emerging market and it is an amazing place to be. Um, I see it as a market that is just taking off. So how, how, how do we get robot arms out of the factory? How do we do things with robot arms that are not just canned routines that you execute over and over again? Um, and where is Rosie robot really? Where's the, the general robot, the robot that just does, um, that can do general tasks. Uh, right now it seems that robots are, are mostly used in highly structured environments. Um, or, and we want to get them beyond highly structured. So what's preventing it's harbor costs. Harder is really expensive comes to robotics. Uh, software capabilities in many cases, the software, um, I mean, it's just a really new market and the software isn't very mature in a lot of ways. Uh, safety concerns, it's, it's not just concerns. It's, it's, there are real issues with safety and robotics and it takes a lot of work to make a robot, uh, safe to operate around humans. Uh, robot mobility, um, humans are incredibly dexterous. And when you try to, uh, do some tasks that humans do, uh, almost intuitively with a robot, you regularly run into all these situations you hadn't thought of. Um, a real basic example is just joint constraints for certain types of movements. Um, and then lack of standardization. There's a lot of people doing their own thing. So, uh, go into a little bit more detail of those various things, harbor costs, um, there are ways we can get the costs down economies of scale. Um, but one of the big problems is like for robotics to be really useful. Oftentimes they require really high precision parts and those are expensive and difficult to manufacture and then differ, then you have to calibrate them. Um, and you have to have software that can account for all the, uh, mechanical attributes of the part. It's, it's much more complicated than a lot of mechanical devices. Um, and so, uh, and in the end, humans have great closely control. Humans are actually really good and dexterous at doing tasks and building mechanical systems to do them is, is hard. Um, software, um, it's, software is not well designed to deal with things like, partial occlusions and cluttered environments and nonlinear motions and, um, reasoning about unpredictable situations is, is, is still really hard and kind of an unsolved problem. So, um, safety, um, these are some like simple examples are easy to think of more. Um, and there are, in many cases you can create structured environments. Um, you can modify the environment in a way where the robot works to mitigate safety. Um, and that's one of the really common approaches. Another common approach is building robots that are just kind of inherently safe and the robot vacuum is a great example of something that's just kind of account or inherently safe. Like the way it's designed, it, it like mechanically can't do much harm. So that means that you don't have to worry about this one as much in that case. That's a great example for ability. Um, go back to the robot vacuum, uh, navigating stairs is not a thing that anybody's robot vacuum does. Um, and yet many people's homes and apartments have stairs in them. Um, humans are really good at navigating stairs. Stairs were designed for humans, not for, not really for a wheeled robot. So there are like robots there, uh, mentally more complicated and more, and more costly. Um, so, yeah, there's all sorts of reasons. Um, lack of standardization. So every robot manufacturer seems to have their own, um, software standardize. Uh, software standard and reminding that the XKCD joke where like everybody's like, we'll create another standard to unify them all and you just have one more standard. Um, the reason they do this is for vendor lock in, but it really, in our perspective, it's really slowed the, um, the progress of the technology and the real cost here. So why do current solutions not address the Rosie problem? I've been giving you some, but go more specific, uh, off the shelf hardware. Um, the programming environments for off the shelf hardware is really restrictive, right? They're often designed around specific tasks that the creator of the hardware, like envision the robot doing like currently the area where robots are really, um, successful or arm robots are really successful is in these repetitive manufacturing tasks. So a lot of the office shelf hardware is really designed around doing repetitive, um, tasks. And so the programming environments are just designed for helping you do that. They don't, they don't enable general solutions. Um, and then, yeah, software licensing. I mean, we're, uh, we're talking about, um, open source software here and software licensing is a big problem with standardization. So, uh, existing software frameworks, um, there's a bunch of them. Uh, they, going back to this thing, they, uh, they don't offer much flexibility and the creators of them envision like the current applications for robots as their only target customers. Um, and they're just difficult to work with. So yeah, this is just more of the, like from the hardware side, the hardware is really designed around industrial reliability and robustness and, um, the needs of non experts to be able to like make canned motions. Um, and so it's really hard to use a lot of the current hardware and software, um, to do more general things. So this is where we're at where it's, it's, um, because robots are designed for the end user and the end user really being factories. Um, it's hard to use them for, uh, novel tasks. So an alternative brush, this is, this is really just the pitch of like who we are as picnic and what are we doing about this? Um, we're involved with Ross. You've been hearing a lot from, uh, other people who are involved in Ross. It's, um, it's a open source standard with a good license. Um, for robotic software and there's no vendor lock in and it, it offers interoperability like the different, the different components you use in building software is oftentimes very hardware agnostics. It solves a very specific problem and it's composable. And it's really built around these sort of developing open standards for the software that everybody seems to write. Um, yeah. So robust general purpose robot hard software is hard to write. And it, it means that, um, really one of the best ways to achieve this general purpose robot software is with a lot of input from many different people in, in different industries, even who, um, bring their problems and help develop a general solution that is, um, works well. It's sort of the beauty of the open source model in a lot of ways. Um, why should you pay attention to Ross? Well, it reduces cost because a lot of the work is already done. Um, it's easier to hire talent to work on your projects if you're using Ross because it is developed enough of a life of its own that there, if you need people to build robot software, there's a lot of people who already really like Ross and already really understand it. And there's an active community to support people and, and to find people in. Um, it helps with tech transfers from academia. Uh, there's a lot of people who are doing research who are using Ross because in the same way that enables innovation from, in companies, it enables innovation and research. Um, and it's built on top of industry standards. The latest version of Ross, Ross, too, is, um, built on middleware. It's based on DDS, which is they, um, an industry standard for, uh, middleware. So big companies are standardizing on Ross. Um, there's a really impressive steering committee. Um, we are honored to be part of it. Um, and then there's several consortias that have been established. Also Ross industrial that really does deal with industrial applications. Ross, um, auto aware, which is the, uh, people are building self-driving car like things based on Ross. And then the movie community, which is us. Um, so how Ross is helping us achieve the next level of robotic capabilities. Um, it's helping commoditize hardware in that it defines abstractions for hardware and interfaces to hardware that are standard. Um, and allows you to write software that's agnostic of hardware and can compare hardware. And then once you actually get hardware, or if you change what hardware you're using, it's easy to swap out. Um, there's just more software capability. And it's all, um, nicely integrated. There's a robot mobility mobility. So navigation to is another huge project in the Ross world. And, um, that really helps with mobile robots. Um, and it solves the lack of standardization problem. Um, in that Ross is a open standard in a lot of ways. And it's, it's, it's really pretty straightforward and universal. Um, except it's accelerating robotics innovation. Right. Uh, there's an image from fetch. There's hundreds of startups. Um, there's a lot of research papers. There's hundreds of millions of dollars in company evaluations that are based on Ross. Um, so when is this not the solution for you? Like why would you like what, in what cases would you not, would you not follow this? If you, if you don't need to customize things very well, or like you don't, you don't have any really specific needs. If the current solutions meet your needs really well. Um, you're not building a robot product yourself. Um, and, uh, you don't care about owning the IP and then lastly, you don't have an expert development team. One of the real costs of, uh, developing one of the hardest costs to feel with the developing robotic applications. Um, that are novel is that you do need a really, uh, professional development team. Um, may I ask what about my company's IP? Right. You want to build something, um, and sell it and you want to, you're concerned about protecting your IP. Um, you don't have to give it away. Um, non-competitive contributions are greatly encouraged. It's, it's encouraged that companies that are building stuff on top of Ross when they, um, that they upstream as much as they can. Um, Ross is not copy left. So you can build proprietary software on top of Ross. Um, you can build a, uh, a fork of Ross itself if you wanted to and sell it. Um, the. Yeah. You, you can focus on just like what your core value is, what your competitive advantages and not on having to build all the infrastructure because a lot of it's done for you if you use Ross. Um, so what is move it? Move it is the thing that I'm here to talk to you about. I give you this introduction to Ross robotics. Um, it's a motion planning platform. Um, it started as a project called arm navigation and then changed names. Over time and now we're on move it to move it to really just signifies like move it that works on Ross too. Um, and that's, that's getting much more mature. It's, it's several years old now. And, um, it's where a lot of the new development is happening. Um, it's incredible how, how, how long it's kind of thing. Um, what, what, what does move it help you with? Move it is really about, uh, robot arm manipulation. It helps with motion planning. Um, it combines all sorts of different kinds of planners. So in, in an abstract way. So we talked about hardware as an abstraction, right? You can swap out hardware and it's, it's straightforward. Well, you can swap out planners with move it. It's pretty straightforward. Um, a bunch of different collision checkers. Um, inverse kinematics manipulation, grasping control, 3d perception. Um, this is sort of in the order of like things it does really well to the, the things that, um, use more work. Um, it has a really feature rich ecosystem. We, we work with a whole bunch of different global planners and Cartesian planners and inverse kinematics libraries and grasping libraries and, um, a couple of different collision checkers. And, and then also with perception and octomaps, right? I said, this, this is oftentimes when you get into this space of not just doing a canned motion. Um, one thing that's implied is that you're reacting to a changing environment and that means some sort of perception system. And that perception system needs to feed into your collision environment so that you can plan within that environment. So move it has tools for that. Why Rust 2? Um, Rust 2 is designed for production. Um, it shortens your time to market in a lot of ways. Um, Rust 1 really, uh, in a lot of ways, Rust 2 really is, it's the authors of Rust 1, um, got to, uh, a lot of people started using Rust for industrial applications. They had a lot of feedback and a lot of that went into, uh, redesign of some of the core pieces. And that's what Rust 2 is on its multi-platform. It works on Linux, Windows, Mac OS. Linux, really, while, um, Linux and Windows are, um, tier one supported and Microsoft is putting a lot of money into making Windows work better. Um, it is still, uh, the only real first-class, uh, Rust environment as Linux. Um, and then now because of how they've abstracted executors and various other things, uh, you can do more, um, hard and soft real-time stuff. So, let's talk about...