A. Schreiber

Musings of an obsessive robot builder.

[aschreiber]

On Feedback

I know it looks like I took 2 weeks to mount encoders but I didn't. I spent about 15 minutes last night soldering the connectors onto the CTRE Encoder breakouts, plugged the wires in, and then had readings.

This really gets into one of my rants about FRC - Encoders and feedback systems are always an after thought left up to the implementation. The issue with this is that feedback devices typically need to be mounted accurately and many lower resource teams can't hold the tolerances needed to correctly use them. The very nice part about the Rev motors I'm using are that they have encoders built directly onto them, in fact, you can't get them without encoders. Feedback by default should be the norm. The VexPro versaplantaries were a good start in this direction but need to come as either a Gadgeteer port or a 4 pin molex connector. Additionally, the Armabot encoders for the VP 775Pros are a solid start as well but because they are a separate purchase many teams won't buy them or design them in from the start.

Either way, I've spent the last 2 weeks driving the chassis around to see if anything major breaks. The good news - it doesn't seem to. Handles well, is very controllable and predictable. The only thing I wish it had was a bit more speed. Revision 2 I guess.

Next steps now that encoders are mounted and reading is drive a set distance and measure repeatability. Source at this point is available at https://github.com/schreiaj/HeroBot/tree/v1.0

[aschreiber]

Got the chassis running last night. The noise is because the Rev HD Hex motors seem to be quite noisy. But I imagine the timing belts running on the 3d printed pulleys may not be the most efficient thing ever.

This consists of a CTRE Hero board, 2 CTRE Talon SRX motor controllers, and 2 Rev Robotics HD Hex motors. It's all programmed using C# and powered by a 2200 mAh 3S Lipo.

Also, Bear does not enjoy the new robot overlords.

[aschreiber]

Been busy assembling the new printer. Instructions have been ok. There's a few steps that are very full of WTF but overall it's been pleasant.

[aschreiber]

Starting Hero Development

So, while waiting for the CTRE Hero board I ordered I downloaded Visual Studio and the Hero SDK and read all the documentation. I then decided to implement what has colloquially become known as Cheesy Drive in the FRC world. For folks in the real world- this is split arcade. One stick controls throttle and the other controls steering. Think of it as how you would control a character in Halo.

Not going to lie - it was fairly simple but I learned how to access Talon SRX's and the USB gamepad as well as remembered how to write C# code.

All said, a fairly nice experience.

[aschreiber]

Ok, so it's been a really long time since I updated anything here. Since then I've bought a house, sold it, moved back to Tampa, sold my Replicator 2X, bought a Prusa Mk2, learned how to use a laser cutter, built a small driving chassis with it, adopted 3 cats and had to put one of them down.

Now that you're all caught up here's the big news. I'm quitting mentoring FRC. I no longer have the patience to deal with students on a day to day basis. I will still be volunteering at events as a judge (this year it's looking like just the Houston Championship) But more importantly I'll be starting doing a lot more development of new tools and approaches. That's where the laser cutter and small driving chassis come in. Until recently it was exceedingly expensive to build and test any sort of drivable chassis. Thanks to REV Robotics releasing a new range of FTC targetted parts it became cost effective to build an FTC scale chassis to do software development on. For the time being I have ordered a pair of Talon SRX speed controllers, a CTRE Hero controller board, and the appropriate cables and adapters needed to read the encoders installed in the chassis. The electronics guts will be here shortly. Check back for more details.

I plan to publish parts lists and design files for the chassis so anyone can replicate it. The chassis was originally designed using the Educational version of OnShape so sharing that will be rather easy. I am planning on updating to use more available parts and be cleaner.

[aschreiber]

2014 Gifts for Developing STEM in Kids

So, I've had a handful of folks ask me for suggestions on what to get their kids who are interested in technology or who they want to get interested in technology. So, I'm tossing together a quick post to give folks some ideas. This is NOT a product review, in fact, it's not even based on me owning or using the products in some cases, mostly based on knowledge of the product through seeing it or reading actual reviews. Take this as a jumping off point to conduct your own reviews. Most of these suggestions are for parents, they range from $50 up to $400 in price. Just keep in mind, that most of these will require some technical know how from the parents to support the kid.

LEGO Mindstorms Kit

349.95 - Amazon Suggested Age 8-14

This is the big obvious one. Most children have LEGO, most children love robots. The predecessor to this kit was what got me started 15 years ago with robotics and it's only gotten better. The new version of this kit connects to Wifi and bluetooth, is remotely controllable from smartphones, and is fully programmable using included software. In short, if you want to kickstart an interest in robotics, this is the most common way to do it. Not technical yourself? Fret not, there's a plethora of books and websites out there dedicated to fixing issues.

Ok, yes, it's expensive. But it's only slightly more than an XBox One. But I get that, I'd suggest looking on EBay or Craigslist to find a used one or the older edition. While the EV3 is faster most kids won't ever push the limits of the older version.

VEX IQ Super Kit

299.99 Suggested Age 8-14

Ok, I'm hesitant to put this on here, it's similar in functionality to the Mindstorms kit above (programming, building, etc) but it has a slightly more Erector Set quality to it. As a building system it also has slightly more advanced parts such as omni directional wheels and a wide range of sensors. The reason I'm hesitant to add it is because it's actually NOT available in time for the holidays. It ships in 8 weeks, so maybe a birthday gift. The reason I have to include it is because it is coupled with a full training curriculum.

There's also the bigger VEX Robotics kit, it clocks in even pricier at $499 plus a license for the programming tool and the actual programming hardware. It's really pricey. Solid, extensible, and real robotics tools.

Sphero

$99 Suggested Age 6-14

Ok, a bit cheaper on this one. This is more of a toy, it's a bluetooth controlled robot that kids can control from their smart devices (or yours). It's water proof, it's drop proof, it's fun. But want to know the real fun part? It's fully programmable in a variety of languages. There's a ton of apps out there for doing different things with it. It's not terribly extensible but if you've got young kids who are rough on toys, this might be a good one for you.

SFE RedBot

$74.95 Suggested Age 12-16

Up until now the requisite technical knowhow has all been included or easily available. Here's where we go off the rails a bit. This is a robot kit. It requires some basic tools to get started (ok, it requires a screwdriver and that's included). No soldering is needed for this kit, everything is plug and play. But the kit is useless unless it is programmed. The good news, the Arduino environment is considered quite beginner friendly and there's a lot of support. If your child is interested in programming this is probably one of the cheapest mobile kits out there for it. (I looked several years ago, the only ones I could find reasonable prices on were some chassis from China with no drivers and the documentation was in Chinese. These are actually cheaper than that!) As an aside, I trust Sparkfun to stand by their product, if you have issues contacting them will likely resolve them.

Those are some of the suggestions I have made. Don't forget there's always the options of purchasing a Raspberry Pi and installing Minecraft on it , playing with Arduinos, or maybe even spending some time learning to program together. And you know what? If you don't feel comfortable doing this stuff alone, find your local hackerspace/makerspace and I'm sure there'll be dozens of folks who would love to help you.

Happy Holidays.

[aschreiber]

Printing a frog for the GF. Why a frog? Dunno. She wanted it

[aschreiber]

Robots...

So, I haven't been posting lately. Why? Robots. Truth, most of what I've been doing lately has been in support of that. So I've made a lot of software and what not. The good news, it's all open sourced because that's how I roll! Yo

Interesting stuff I posted yesterday, https://gist.github.com/schreiaj/8592755 So, what this does is generates a trapezoidal motion profile for your system to follow. Why is this important? It helps you get more repeatable motion. Useful for autonomous motion. Would likely be useful for rotary motion too but I'm not doing that because we don't need it.

This is a really clever implementation courtesy of Paul Copioli's posts on CD about the topic (I closed the tab, maybe I'll find it later). Tonight I'm going to work on moving it to Java so we can use it on the robot. Really I was just curious how the implementation worked.

The easy way to use this is to modify the values in the script for your scenario. It'll spit out TSV data. You can use this in excel if you want (copy pasta) and plot it there. OR you can do it the easy way using gnuplot. I wrote a quick gnuplot script for it this morning:

set terminal png set name "test.png" plot '< cat' using [1:2] with lines

I named it something (tsv_plot.script) and then get a "nice" plot using simply:

ruby trapezoid.rb | gnuplot tsv_plot.script && open test.png

[aschreiber]

I was bored over Christmas so decided to see if I could print some simple robot parts. These are the Versa Blocks from Vex Pro. So I MAY have started printing out enough for a rolling DT. Call it a test to see what sort of load 3d printed parts can deal with. Honestly, if I were trying to save some money I likely wouldn't bother printing these parts though, a set takes just shy of 2 hours of print time. It's not BAD if they hold up but there's likely better uses of 2 hours of printer time. Plus, I'm not sure they'll hold up.

Oh, and also, my parents love the 3d printer. Now if only I could find a gal who was as nerdy as I am.

[aschreiber]

Update

Well, it's been a while since I blogged. Honestly, it's been a while since I've done much of anything outside of work. The main reason is that I'm actually moving. Currently the majority of my stuff is packed away, I've got the parts for Caelus with me and Vaul is set up on a table at my parent's place. Otherwise, I'm sorta homeless, hopefully I can take advantage of this to finish Caelus but idk where I'm going to fly him, the weather here is a balmy 9 degrees, there's 1/8" of ice on everything, and a nice dusting of snow. Maybe I can find a warehouse...

[aschreiber]

Apparently the Lions thought winning is more important than thanksgiving tradition

Chris

[aschreiber]

Caelus Update

So, for an update - all but one of the ESCs is soldered onto the PDB and has their mating bullet connectors attached. Thanks to Sean for helping me with that, I really suck at soldering.

We also replaced the strange connectors HK had on the batteries with some Dean's T connectors. With only minor fireworks - remember kids, don't cut through the leads of a battery at the same time :\ OOPS

I've got today, and tomorrow off then the weekend. Still to do is print a mounting plate for the controller board, route power to it, and print holders for the bullet connectors to get zip tied to the frame. I guess that means I need to pull Vaul out of his packing.

But today, if you'll excuse me - I have to go make pasta, watch the Lions lose, laugh at people in the north while I still can.

[aschreiber]

This video explains why I'm building Caelus. Well, that and I just wanted to.

[aschreiber]

Documentation

So, one of the biggest issues I deal with when doing a project is documenting it. I'm just terrible at taking pictures/making microcommits. With the hardware projects (like Caelus) I am fortunate in that my parents are crazy interested in it and it's far easier to write a blog post and drop them a link to it. Software stuff is far harder for me thus why not a lot of stuff shows up on here.

Right now I'm struggling with figuring out how best to cover builds like the quadcopter. On one hand I can do in progress posts but then anyone trying to recreate the work has to jump through a ton of posts. On the other, I could do monolithic posts which forces me to WRITE monolithic posts which I don't do well. Seriously, most posts on here are less than 2 paragraphs for a reason... I'm bad at long form writing and tend to ramble.

We'll see, heck, I barely know who reads this stuff. I really need to install Google Analytics.

[aschreiber]

Is there anything quite like wandering a hobby store searching hundreds of poorly labeled packets for the one you need? Yes, it's called dating.

[aschreiber]

So, Caelus is mostly assembled as a result of a marathon session of mounting props. Turns out he's mechanically quite simple. Still to do is purchase 3.5 mm bullet connectors for the ESCs, print a mount for the controller board (2mm thick?) and getting everything programmed. I've got a class on "everything quadcopter" Thursday so this might work out well :)

Edit - I've gotten some flak for having a finger in the picture... That person is a minor, didn't want to show his face.

[aschreiber]

Why I haven't been posting lately.