Design of the LoliBot hardware

[Music] [Music] so thank you everyone thank you Andy so I'm gonna try to keep this talk pretty high-level I've got a bunch of different topics so I sort of want to cover and they're sort of vaguely connected but I'm not gonna get in to anything deep because I find that we tend to just leave people behind a little bit public speaking is one of my special skills so you're in for a treat this is gonna be fun it's gonna be rambling poorly written and poorly timed so and the jokes will likely only be funny to me so let's here we go first I just thought I'd cover with what actually lowly buddy is I haven't really told you what I'm gonna be talking about just gonna have to come with the riot the lolly bot is a ping pong soccer playing robot at the moment it doesn't play ping pong yet so that's exercise for the reader I guess it's run on a roll in 32 light it has two wheels six a six axis accelerometer a little servo to kick balls around and it has an IR sensor on the front I thought I'd introduced the team pretty early before I get lost a little bit we've got John Aqsa who is missing ah we've got Andy kill me Angus Grandin who's not here today unfortunately but as sp30 to support agent for our entire team he does a pretty good job I've got Nick Moore who I haven't seen in here yet either by myself because I always forget to mention myself we've got Thomas helping out Steve helping out Kerry helping out and John helping out and other helpers who I have missed I'm sorry guys I didn't have a complete list before the hack Melbourne guys in Melbourne I've done a great deal to make happen as well just literally the day before we flew up we had one of our guys come out and do all of the tagging and testing for our equipment and I'd like to make a special call-out to mark Melman not so much for this year's project yet but for last year's project where he went back and completely wrote all of the firmware we had and I think that deserves a special mention thanks mark all right what did I talk about next ah the open hardware mini conf we don't normally talk about what that open hardware mini confer is about I believe it's the ninth year although I think the website says eight years Andy says nine years it started as the Arduino mini cough and each year we try and design a bit of a custom board that's a bit new bit of sort of bleeding edge tech but there's also simple enough that you guys can put it together on the 9th year thing I've just realized I do I've done four years of this so I'm nearly halfway which is not a bad effort all right John's done all of them and he's done bit over half I would've thought so I thought I'd talk also about our design methodology and our timelines for the open hardware mini cough so in January we have the conference we get very excited because we've finished it we're through the hump high on success we have fantastic ideas about what we'll do the next year we also promise ourselves that we won't leave it so late next time and we will get ahead of the curve around Maish we have a call for papers we email each other nothing really happens maybe some more ideas get thrown around we also promised that we won't leave it so late and we will stay ahead of this curve July there's the deadline ish for the papers we actually meet up we have an idea we Norma sort of vote on it we relax a decision has been made we don't need to worry we also promise that we won't leave it so late September someone remembers that we actually have to do the project so we design a prototype the idea October we spin the hardware and inevitably it does not work November we saw a second spin and it mostly works you all have that what everyone - lolly bot has a little that doesn't quite fit into the slot that it has provided December there's normally a cut-off when we can order components and expect to get them and the boards normally take a little while if we have to do any surface mount stuff overseas that takes up probably three to four weeks longer and then of course early January we frantically fix patch bits that don't work we write code we panic we do some more panicking we end up in the pit of despair and Annie has no sleep for days that's been the last two or three years and of course late January we have successful events everyone is having a great time it's a very rewarding experience we go home and go back to step one that's been the last three years that I've done it and I suspect it's been oddly used as well so this year we made a few design decisions the SP 32 chips they are still great they have bluetooth they have good power saving stuff now you can actually get dev boards when we made last year's project dev boards just didn't exist at one point we had 40 of these in a box and hacker space and I think it was the the largest some of these boards in existence in the entire world we decided to try and stop shaving yaks a bit so the idea of yak shaving of course is that you want to make a sweater and you end up growing your yak and shaving it so we've based this year's project on an existing board that already does most of the stuff we want it to do everyone still seems to like doing a bit of soldering that's great and theoretically the simpler hardware means a better opportunity for software we did have a bit of gold plating we added expansion headers so they're there if you teach him extra pins we have an IR sensor wasn't in the original spec but it's there now and we had LEDs oh there we go it spins around it's great we had some gold plating the didn't stick so we did actually throw some stuff out we didn't just toss everything on we originally wanted to plan when we plan to use a remote control with the e badge the e badge was a bit of a bridge too far and the timelines we had and we actually did get ahead of the ball on that one which slowed up our production of the lolly bought a little bit I suspect maybe next year we had a boost converter because we wanted to up the amount of voltage to the wheels because more voltage means more speed in this case that added a lot to the build difficulties so we thought we'd leave that off and we made another board called the Lally wrapper it plugs into those expansion headers the designs complete bug John about it maybe we'll get some extra ones but yeah they should be fun ah controller software I've only put a little bit here because Andy's gonna be doing a big talk about this thing but micro Python is really quite mature now we've we've been building on stacks that the team preaching clans have already been using MQTT is great for rapid prototyping I have no idea what's going to happen if you all get it running at once I suspect something will melt unfortunately Bluetooth on the SPT tastes a little bit flaky so we've left that off for the moment and that's me finishing talking about the lolly board now this bit might get a bit weird but let's talk a little bit about circuit design because we we always seem to do that and I'm sorry introducing something here which I don't think anyone's seen it's a bit of a new product I don't know how polished it is it's a little apt I've gone up one it's a little app called concept CAD gives you some idea on how your program stuff and here you can sort of see you can do some circuit board layout and by now you should have worked out that this is actually a game so the reason I've included this is because particularly for me when I'm doing circuit design this sort of this is the itch it scratches in my head I could have been doing this in fact I think last year this probably took a few hours out of the design work playing this particular game but you could also spend it doing something more productive so I thought I'd show a little bit of a comparison between games and real life so in the game you can rate yourself against your friends which is great because that one I just showed you there one of my friends has made his like one power unit smarter than mine and I have no idea how we did it virtual circuits are really cheap if you can do something virtually it saves you a lot of particular if you like me and it takes you two or three but also get fired in your for your in-game job for spending too much time playing games and it's actually a really good solitaire game which doesn't help on the other side circuits actually do something real circuits actually do something you might not know exactly what that is until you get it but it will do something if you're building at home you get to play with chemicals I don't know if that's a positive or not that's up to you and you get an actual circuit to love and to hold and you can show it off to a whole room of your soon-to-be friends and friends from previous years so that's a big plus as well if you actually really want to design stuff I've been using KY CAD for three plus years now I picked it up at the hacker space I'm not going to dig into it really deep it's open source it's fairly mature it's free as in beer and speech and it's filled with just as many quirks as every other CAD tool that you're used to using so you really need to check your fin your pin footprints because I think there's a reason why the plus and the minus on your your robot feels a bit wrong that's because the switch between versions oh yes it has a schematic editor as you'd expect it has a PCB editor which is what you take your schematic editor and then you push all the wires around so you can what's in front of you that's the design you have in front of you and it's got a lovely little 3d viewer which hey if you're trying to sell something to the rest of the guys that you you're working with that's that'll get you there so that was my quick kai CAD talk now I'm going to go segue again and talk about sort of open hardware in general in particular how I approach with how I build another project with using Chi CAD plus other tools to produce an end result that is open and real hard well I say we hide what that's not right like it's got physical elements to it beyond just a circuit board and that is the auto changer this is a project for my wife's kneading machine which she's going to be talking about in depth tomorrow but I'm going to cover that Auto change apart here because that's uh that's something I've spent the last six months on sort of held up a little bit with open hardware cough but still you might be asking what that is if you've never seen no a kneading machine before mini machines Wendy multicolored there's certain different ways to do it but a manual device is used to switch between them you can see here on this video that every pass of this you have to hit a button you can't automate these machines to a bit with the robotic arm but you still have to stand there and go click click click it slows you down gives you a nice cadence but it's fairly annoying if you're sitting behind it like I am because my desk is directly behind that kneading machine not that I made that any better but we can at least make something that moves on its own right so this is my design which is not playing have a second this is the first iteration of my design servo is pushing little levers I'll dig into this a little bit more and sort of so the different levels I had that one's powered by just a how do we know you know whoops there we go first of course there's the brains of the unit I thought I'd go through that independently of the hardware that the physical bit that sits around it first prototype was an Arduino Uno with a serve a shield and a special call-out to hot glue because that's 90 percent of what it's actually made of server she was really not needed because it only drives for servos has a beeper that yeah like I said I thought the clicker was annoying I I was wrong and then I built something that was worse so that's probably not ideal but you need a bit of an audio cue to say that you've already gone past the point where you need to change colors prototype 2 was a custom board an Arduino compatible using an Arduino mega three to 8p and serial wire CH 340 I did a few spins of that but I always found it was fair a little bit unreliable probably due to my own well I sell Linux does that mean you don't really don't go to university you just how to make circuits so sort of stabbing in the dark a little bit maybe they did a prototype three which is a custom board at it just uses an Arduino Nano that plugs in the bottom and I found that roughly the pricing for was just buying a nano was about the same as buying the chg 340 the 80 mega the crystals and all the extra stuff I needed to make it work with the added bonus that you just pull it off and replace it so that's sort of I think how a lot of this sort of circuit board design goes you try things you find something that works better and then you just replace now talk about the mechanical bits I designed it in Sketchup which and I used another program called Kanban which are not open source they're not well they're kind of free not really though I cut these out on the see th laser see chases laser cutter which is using marlin which is a 3d printer software and controlled using products which are free and I've released the design free so I think I get a technical pass on that one I did about three iterations this was my first interation using a lovely Sketchup render that's how it ended up it does you might know it looks nothing at all like that render this render if you look at those gears they were great to draw they looked really fantastic they are tiny you you get almost no grip on those the laser cutter just went yeah that's just a circle good luck circles don't move levers so I switched for the the levers from the servos and and a bit of pins and then there's that hot glue I mentioned which holds it all together cable ties one of the arms was a bit short so just extended that out that's why we prototype things so I moved on to iteration two which I don't have a design pic for you see it's got much the same design that still these speak square purple pieces which are the arms that move out the gears much bigger chunkier easy to cut on the laser cutter move about same speed the more leverage you get any gear obviously the faster it moves and this was not a bad compromise designed three minor changes just to improve the construction of it you see those arms are thinner but there's more mass on the base which is what actually provides all the strength the sideways motion it's easier to assemble still uses the same sort of gearing approach the arms are extended little fancy bits like the little curve around the end that just helps the the yarn feed through and the reason I don't have a physical picture of that one is because this is what it currently looks like sitting on my desk that was about the same time as we hit crunch for open hardware mini-com so maybe one day I'll finish it and that's bordering on becoming a worthwhile project to sell to the other knitting machine leaders that are out there so what does it actually look like when it's running I don't think you can hear music on that one but it's actually fairly annoying how noisy it is but that's in demo mode and how quickly it can sort of flick between and change not that you'd ever be able to move that quickly with the knitting machine it's got a little infrared sensor at the end which tells when the characters moved past and I'll move to a picture in a second that just sort of shows how that goes which is this one this is the this is revision two because obviously I have a future revision three that's with the robotic arm which you cut which if you're doing that with the manual you sit there the whole time go and click click click click click so yeah moves fairly quickly fairly happy with the approach 15 minutes to go that's a good call at probably not going to happen and those are the needles that move in and out as I said my wife's talking about this in more detail tomorrow she has a lot more videos of how a kneading machine actually works but I just wanted to run people through this idea of how how you can take it an idea and push it out into a larger design I will point out that the first time she asked me if I could make something like this for I took one look at it and said there's this big metal arm thing I've got no chance of doing that if you just think a little bit closer about how a problem is solved you can actually get a long way with some fairly simple tools and last I'm pretty much done thanks for listening I hope that wasn't as bad as it could have been here's all my links they're all at the end where it's easy to get to and like I said please drop in and see Sierra's talk last time we got she's got 45 minutes I think we were down to 52 minutes yesterday so she might make it under time thank you for doing the holy book project and I think I'm done any questions any questions part of Pluto thing that's that's just how the branding my wife uses for her all of her kid on the how to Pluto link that's listed there it's got a lot more in depth of how I made that yeah there's a lot of I've really gone in deep on it things there's quite a good one there which shows the different revisions between Sketchup of the version 1 and the version 1.3 I think I call it but that's a little bit yeah anyone else nope excellent let's learn some micro Python we have a quick break where we switch over mics and then Andy will be talking micro Python thank you ah the game that's the that's the third last link there shen zhen AI o is called available on Steam I know nothing like it's puzzle game it's fun

Channel: LinuxConfAu 2018 - Sydney, Australia

Views: 315

Rating: 5 out of 5

Keywords: lca, lca2018, #linux.conf.au#linux#foss#opensource, JohnSpencer

Id: 32KJ72T4gZk

Length: 19min 28sec (1168 seconds)

Published: Sat Feb 10 2018