 And we're here at the Embedded World here in Nuremberg at the Beaglebone Blue Boost. Hello, so who are you? I'm Christine Long. I'm the executive director, CEO here at Beagleboard. And we're excited. We're introducing the Beaglebone Blue. This is a new Beagleboard right here. It's got a very special system in there. It does. It's an open source Linux computer. We're having kind of a focus around robotics and education. So I want to introduce and hand this over to Jason Kreidner, who can speak a little bit more to this. Jason? And who are the other guys here? So we have Jason. These are board members of Beagleboard. Some of our board members. Who? Jason Kreidner. Who are you? I'm Drew Fustini. I'm one of the board members of the Beagleboard at Org Foundation. I'm actually a big fan of your videos. I've seen you on Google+, right? Yes, yes. I'm on Google+. And you were just in Budapest, right? Yeah. I was watching your LinaroConnect videos. There's a lot more videos coming from the crazy Linaro engineers doing all kinds of stuff. Yes. And I've gotten some messages that I really have to check out the Beagleboard Blue. Beaglebone Blue. Beaglebone Blue. And so what's going on with that blue t-shirt and everything? Hello, who are you? I'm Kepp Radford. I'm also a board member of the Beagleboard Foundation. All right. And you're holding a special implementation right here? I'm holding a little robot that's an EduMip+. Well, we're going to check it out in a second. And hi, who are you? I'm Jason Kreidner. So I'm a co-founder of Beagleboard.org and one of the board members as well. So what are we looking at here? Let's check this board. So this is the new board from Beagleboard.org. This is the Beaglebone Blue. It's available today from Arrow Element 14 and Mouser. The suggested retail is around $79 for this. And this is a 1 GHz Linux computer based on the Beaglebone Black Wireless, but integrating all of the stuff you need for robotics. So it's got motor drivers, it's got quadrature encoder inputs, it's got sensors, it's got a 9-axis IMU, barometer, essentially everything you need. You have the previous one, the black? The black? Oh, okay. If you can hold it right here in the light. So you say, what is this? This is not an SOC. This is a system and package. So this is using the Octavo Systems OSD 3358, so they take about 154 components off the Beaglebone Black. So 154 from this one? Yes. What kind of components? So the processor, the power management and the DDR. The DDR is especially interesting because it takes all the high speed stuff that you'd have to lay out on your board and puts it inside the one chip so that you don't have to lay it out yourself. So this was all done in a four-layer design in Autodesk Eagle. So it's something that's really affordable for individual hackers that they can go and make their own boards. In fact, we released, on our GitHub, we've released the design for the Beaglebone Blue already so that you can actually go and look at that design and make a modification so already. So by putting all this stuff into the SIP, you know, doing stuff, you know, with What is that? Love more affordable tools. So this is the inside of the SIP, right? So this is, you know, a special epoxy version just to kind of see what all is inside of there, right? So you've got the four ICs, two-for-power management, and the processor. And then the DDR memory is in there. And then there's 150 passes, including the inductors, the coupling capacitors, everything that you need to build a system. This one right here, so what does it say? This is 1 GHz ARM Cortex-A8. Yes, 512 megs of RAM, 4 gigabytes of flash with Debian Linux installed on it. And you've got the Wi-Fi and Bluetooth built in, and the Wi-Fi is MIMO, and it provides the ability to run, like, hotspots, right? So you can actually launch multiple, you know, FIs, essentially virtual FIs on the Wi-Fi chip. It comes default configured to act as a hotspot, and you can connect up to that. So this is a very special Octavo, they're a partner in this. Yeah, they're a startup out of Texas. A lot of former TIers is primarily composed of, but they're working with Texas Instruments and Beagleboard closely to do the special packaging design. So they designed the packaging for the SIP that we're using. It's a new way of packaging. It's a new way, right? So they think that, you know, you have to combine a lot of different technologies if you want to make things easy to use. So, you know, they both get the advantage of doing things on a much smaller scale by using the packaging technologies that are used in putting the chips together, putting semiconductors. But the dies themselves don't always blend very well. So the technology that you use for making a processor is different than the silicon technology you use for something like the memories, right? So they're more something like analog, you know, to have, you know, good analog, right? So you're working in blending a lot of different technologies that don't mend themselves well to being on the same die. So SOCs were kind of yesterday, right? They're still very interesting, you know, putting a lot of gates onto the same device, putting all the peripherals, but you're still in kind of one process node, right? That may be really great for doing processors, but not the greatest for analog. So this allows you to put them all into one chip. How does it connect everything together in here? What's the kind of like connectors, I mean, in the... Oh, inside here? So it's using, you know, bond wiring. So you'll see, you'll see in here, and I'm going to get it a little bit better in the light. But you'll see they're soldering down some of the discreet components onto the substrate and as well as using a bond wires to break things off of the individual dies onto the substrate. You can see they've also used, end up using a packaged memory. That's because the memory technology tends to change fairly quickly. And this way they can pull in the latest memory technology. So that's going to allow for a crazy amount of innovation in terms of board designs and products and stuff? Absolutely. Much, much, much higher level of integration into products. And you'll see in this one, I also want to point out there's a lot of empty space on this one. And that looks like it's an accident. It's actually very much on purpose by Octavo. You know, their intention is that if somebody needs to put some extra things into the system, there's an opportunity for them to get die and put that into the system itself. There's got to be expensive, no? To add something in there? It's not so expensive. You know, there has to be a reasonable amount of volume, right? You're not going to make 50 of them. But there's certainly some points where it makes a lot of sense. And for Beagle, it makes sense, right? So it made sense for your project right here. And how's the performance of this compared to previous era kind of products? So just by doing the integration into a SIP, right, we're seeing memory performance improvements. But really, the things about the Beagle bottom blue that make it super exciting are all the robotics interfaces that we put in here and all the work we've done into getting default software. So there's much more, actually on the previous, on the black, right? You ship more than a million, right? Yes. So more than a million, that's like a huge success in the development board. Absolutely, absolutely. I think we've definitely done some major things in the space. And all kinds of people are using it, makers, but schools. Lots of professionals. We're here at Embedded World. We're talking to a lot of real customers that are actually doing things, embedding into products. And by being open hardware, they can go and make their own supplier agreements and do their own sourcing. And if you just compare it. So what we do is a small portion of what people actually, of all the things that people do in derived designs. And there you have on the black, you have those connectors right here. Yeah, those are our K-Petter connectors. You see on the blue, those have actually been removed. It took them away. And yeah, and it's not so much removed. It's like, well, we've done, we took the Beaglebone black wireless and we integrated a robotics cape. So with all those extra sensors, the battery management. The wireless, yeah. Right. Well, so this is the black, this is the black wireless. We launched this back in October. So this is our, you know, up until now, this is our newest board. Yeah. And what's the reception being on that one? It's been very good. Yeah, no, it's been very good. It takes some time, I think, for a lot of people that are doing real product stuff out of it. But for people wanting wireless connectivity and we're using Beaglebone black, it's a pretty natural selection, just as a drop-in. It's a fully cape compatible. Everything just works, right? With the only swap-out being the Wi-Fi for the internet. But I wanted to show you something here extra about the blue. With both the black wireless and the blue, you can program them with National Instruments LabView. And so what you'll see up on the screen is actually, this is actually running a LabView VI on the board and it's network connected to the instrumentation. But the actual application here is running on the board and it's just grabbing the data back. And this is using LabView Home Edition, which is pretty affordable for individual developers. And this is something you can't really do. You can do some of this type of robotics control with a microcontroller. But by running Linux in your robot systems, you have a lot of other instrumentation options, right? Being able to do network streaming and here running high-level applications like LabView. Obviously, you can see the feedback from all the sensors as we're reading off of the IMU. Real time. In real time. Over Bluetooth. This is over the Wi-Fi network, which is why you'll see occasional glitches in the Wi-Fi network. But the actual program is running on the board, right? And then it's just streaming the data over the Wi-Fi, right? It's part of the instrumentation that comes as part of LabView. How much overhead does it take on the system to do the kind of stuff? So we're also running the IDE at the same time. So why don't we look? I don't actually know. Exactly. So this is a web browser. So it's running a web server on the board that hosts the IDE. And so it's currently, the LabView thread is running about half of the processor as it looks right here. So, and there's all these connectors right here. Can you describe, because those, you didn't have them in the black before, right? So what can people do? You say it's robotics? Yeah, so we've got, I'm gonna go ahead and break up some of this to the, back to the IDE. Because it allows me to show you some more of the things on here. So we've got DC motor drivers. So if I type in, into the command prop, I type RC, you see a list of applications that are on there, right? Yeah. So if I run test motors, right? It's gonna give me some R view stack. And I'm just gonna do it with me. All the motors are set, F, T. All right, and if I do this right, yeah, so you'll see the motors turning right now, right? Nice. So there's a C library called Live Robotics Cave. And then there's these ton of example applications that people can start from. And this ships in the distro with it, right? So in addition to the DC motors, there's also quadrature encoders. So if I do a test encoders, try one of those. Test the quadrature encoders. You'll see that as I turn the wheel, right? So as I turn the wheel here, you'll see that the number updates, right? It's to where the position of the wheel is. Nice. So that's using, you know, this encoder board right here to get feedback when I turn the position of the wheel. And you have able to just connect to the right place and do all the stuff that people want to do. That's it, right? Four encoder input ports, four DC motor output ports. We have eight servo motor output ports. Those can also be used to drive, sorry, electronic speed controls after things like quadcopters and hexcopters, right? So an octocopters, right? We have got eight of them, right? You can use the signals to drive those. Those are actually implemented. So we have hardware pulse width modulators on the device. But the servo control here is actually implemented using this PRU microcontrollers, the program of real-time units. You see a lot of people making Beaglebone hacks with. But we come with some standard firmware loads to do that, right? And that means like when things like, if you were to create a crazy bug in your Linux system for your copter, and it was to crash, right? If the A8 stops sending data updates instead of having a free running peripheral, it'll actually detect and time out and cause the quadcopter to gracely fall to the ground rather than to fly off and be a hazard for somebody. So that's the two times 200 megahertz 32 bit PRU? It's programmable real-time units. It's a part of the PRU ICSS, or industrial control subsystem on the AM3358 processor family. It's not an ARM Cortex-R. It's not an ARM Cortex-R. It's a TI proprietary core that's supported by a freely redistributable TI C compiler as well as a fully open source of similar. There's also a GCC implementation for the PRUs. So you can get a fully open source C tool chain if that's what you need. And the one gigahertz ARM Cortex-A8 is a TI? It's a TI, it's part of the AM3358 SOC, which is the one we've been using. It's the same one that we're using on the Beaglebone, right? It's the same processor that we're able to leverage all the same software. It's on the Beaglebone Black. Which has been shipping since 2013. 2013, it was a Beaglebone Black launch, yeah. And here you're showing some examples of drones? Yeah, this is a quadcopter. And we already have support in RGPilot, so which is an open source autopilot code, right? So you can run RGPilot on this today just compiling from the mainline source. They already have support for a blue target. And you can just run that and you can do things like set GPS waypoints, add ultrasonic sensor on here to do, take off and landing support. You can add a GPS for the waypoints. So you have this thing called a mission planner. So that's the open source drone OS? It was originally done for Arduino based autopilot systems. But we did as part of a Google summer code project. We worked with people to port that to Linux. And it became part of the drone code Linux foundation project. And of course the RGPilot project is still around. And so we have support for that project. We have support for blue in RGPilot. So were the black, were people able to do a bunch of drones? They were. How did they do it? So there's several different add-on boards or capes that people can take their BeagleBone Blacks or Black Wirelesss and go and build copters with them. A popular one is BBB Mini. So it provides breakouts to some common sensor boards and drivers to go and build those quadcopters. But now this integrates it all into one, right? It becomes easier, cheaper? It becomes cheaper, for me. Easier, yes, because you don't have to go and connect up a lot of components and debug them. But certainly cheaper because we're buying all these things in volume, right? So if you were to try to piece it all together, it would cost you quite a bit more to build it individually versus just buying BeagleBone. No, it's the same power consumption. I won't claim that. And there you show another drone there. Yes, this is, so for a lot of the robotics development, so the initial robotics cape that was integrated into the BeagleBone Blue was developed at the University of California, San Diego and their coordinated robotics lab. So they've been doing some innovation around UAVs. And this is one of those innovations that they've done, so they've created a hex configuration. You'll notice all the props are tilted slightly. So what this allows them to do is to generate acceleration in any horizontal direction, right? So back, forth, let's write, without having to tilt. So one of the killer applications for the UAVs is with cameras, like doing your selfie cam, like you're doing the types of recording you're doing, you wanna hold the camera level, right? So the typical way a quadcopter would move is it would accelerate one side of the propellers and it would move like this at an angle. And so you'd have to have an expensive gimbal below in order to keep the camera level, right? It's more expensive than the drone. Exactly, the gimbals are more expensive than the drone. And by doing this configuration, they can keep the UAV level as they accelerate in any direction, right? They can generate force in any direction from this. This looks awesome. Sorry, I was interrupting, but the Beagle board has generated how many things? Oh my gosh, I could literally sit here and talk for two weeks straight about all the different things that people have created. You can go online and find videos of where JPL has built Mars rover prototypes and run them under the Polar Ice Caps. You can see things like a DNA thing called open QPCR, which is a, I honestly can spend too much time talking about it. So there's like a million videos basically on YouTube. Absolutely. There's a lot? There's a lot. And there's a whole lot that I can't even talk about, right? I can only talk about the ones that I know I could find openly, right? But really some incredible innovations that have happened. It's easy to get into? It is easy to get into. How does it compare with the Raspberry Pi? Oh, it's just as easy, yeah. There's a lot of examples on Raspberry Pi. And for the most part, you know, it's using Linux. So you just, you could use the exact same samples that you're using for Raspberry Pi for doing Beagle. Arduino is a little bit different in that it's a microcontroller, right? It doesn't have the same performance capability. And in some ways, Arduino's easier to get started with because there's less to be concerned with. In other ways, it's harder to get started with because you don't have the high level interfaces, right? Here, I can just log in, have a command line. For some people, that makes it actually easier to use. So next year there's gonna be like a million drone developers. Absolutely. Right? Absolutely. The kids are gonna be able to start making their own drones? Yeah, much easier, right? Can you 3D print that part? That is 3D printed, right? And so the STL for this will be made public. And so we're working with folks to make sure that these designs are as reproducible by individuals as possible. Or could you mass produce it and sell it for like $5 or something? I don't know about $5. I don't wanna speak for anybody mass producing it. You'll notice this is injection molded, right? So you can get this, you do the same thing, 3D printed for this guy, but you know there's been injection mold today for its frame. Monies, people making all kinds of robots? Yes. So I really feel that for, this is the gateway drug for young people to get into programming electronics. That robots, it's hard to compete with what people do for applications on computers and phones. Then the barrier to entries can seem really, really high. You can see it's pretty easy to get to a point where you could just drive a motor and then it's up to your imagination to actually go and create something. So we see this as very empowering for kids and something to encourage them to go and learn to program, learn about electronics, learn about mechanical systems and controls down the line, right? And get started with something very simple. You'll see we've got some off-the-shelf two-wheel chassis, you see, from Pololu. And, all right, so this is the Pololu Romeo chassis and then this is the Sprout's Runt. So these are off-the-shelf chassis where you can build things out. How much do you cost? So the Pololu Romeo is about $30 for the base with motors and the, so the motors have gearboxes on them and the wheels. So basically everything you need to get a two-wheel rover going for about $30. The encoder boards are like an extra $10 so to get the feedback for the wheel positions. You just have to solder the wires here to the encoder boards. You also need to buy some wires. SparkFun has a large collection of these pre-assembled JST wires. There's a number of other. Pololu also has some of those as well. This is this Sprout Rogue, actually, Runt. Sprout Runt. So this one's a little bit more affordable. It doesn't have the wheel encoders, but I think it's about $18 instead of $30. So it's a little bit more affordable. And some companies are actually releasing mass-produced products, right? Huawei is doing stuff. They started with the, they basically started with Bigelboard and then they were able to switch over to a mode and start doing mass production. Not with your board, but maybe they connect, right? Perhaps even with the board, maybe something like that is possible, but so this is also out of UCSD. So this is using the Adafruit 2.4-inch touchscreen SPI LCD, running LibTFT. So it's stuff that's standardly available in the kernel, a 3D-printed case. There's also a USB webcam. This happens to be a Logitech C270 that's been taken out of its case and then put into a 3D-printed case. There's also, this happens to be a Huawei Bluetooth speaker that's on the back of this to allow you to get some audio output. We can give you a video of this running with the audio, but there's a lot more applications that we have in mind for this, right? You can easily start connecting this up to things like Alexa and providing commands, voice commands for driving your robots, right? What's the idea with the dog? What's the idea? He's cute, right? Yeah. That's the idea. So it's fun to play with, he's always happy. He's always happy. So it's a great community that everybody wants to be a part of and we try to keep it friendly. We try to keep it open and looking forward to getting more developers every day. Nobody knows how smart, potential they really have. I mean, they could be super smart. We just need to be able to communicate with them, right? I don't know what I'm talking about. I don't know what I'm talking about, sorry. That's all right. That sounds great. Right here, people should just start ordering. Yeah, people in blue is available for ordering today. I don't know if there's any stock left, but I would recommend jumping on it pretty quick as people start finding out that this board is available. How do you make more? How quickly can you make more? Oh, we're already making a lot more. We're expecting a lot of demand for this. So there's a lot in the pipeline already.