ROS Robotics Platform

Recently a robotics company called Willow Garage has caught my attention. They have developed three prototypes at this point, one of which I'd seen on various Discovery programs. The Texai robot is a "remote presence system," kind of like a mobile, remote controlled, teleconferencing system. Their PR2 robot is the latest and greatest, and their most ambitious project to date. Think Rosie the robot from the Jetsons cartoon. PR2 even looks something like Rosie, and the project strives to realize the promise of a home robot assistant with social skills. Then there's the robot I'm most excited about, and that's the Turtlebot. The Turtlebot is a development platform for creating knee to waist-high robots that can navigate a home environment. It uses a Microsoft Kinect system as its sensing suite, and a iRobot Create provides the mobility.

First off, it's great that this company is basically taking off-the-shelf components and combining them to quickly produce a fairly robust robotics system at low cost. I have been dreaming about the possibility of having a robot on the first floor of our house which I could access when I'm upstairs or in the basement to see what the dogs are up to, and keep an eye out for my wife coming home from work. It could be connected to a network of fixed sensors around the house to detect activity around the house and go to a nearby window to investigate and record. Maybe the next time someone tries to steal a bike off our porch we could catch them in the act. Or, when the UPS driver actually finds our house for a change we can get a text message that a package has arrived, including a picture out the window of the box. I'll be getting a Turtlebot and building the POLYRO project some point soon, I hope.

But I'm burying the lead, really, about what's so great about Willow Garage. Besides their hardware, what they will be known for will be development of the open source software called ROS, which stands for Robot Operating System, not surprisingly. It comes with libraries of code and applications of particular use to robotics, including visual object tracking and recognition, navigation and object avoidance, and hopefully one day even voice and speech recognition. As an open source project (and such a cool one at that) we can expect to see rapid development and I have every reason to believe it's getting in the game early enough that it could be the major player in home robotics software in the future.

I think it could even be useful to applications besides just mobile, home/office robotics hardware. Why not add a bunch of sensors and controllers to the house and use ROS to control the entire system as if it were a big immobile robot? It's fun to think about anyway.

B-Squares: Modular Solar Power for Gadgets

I'm a backer! B-Squares is a project on Kickstarter, which means that they need a commitment of a certain amount from any number of backers in order to move forward, and backers don't have to pay anything if they don't reach the threshold for funding. It's small-scale venture capitalism.

B-Squares aims to commercially produce hackable, modular solar powered gadgets. It has both an Arduino module and a "proto" module that pretty much has a blank PCB (printed circuit board) in it for you to play with. Through those two modules I think there are a ton of opportunities for neat little things, remote sensing stations in particular. Put them inside big pickle jars on poles and you can have a pretty much autonomous outposts in the wild.
I went in for $100 to get the developer kit when it comes out. I'm really excited about the possibility of powering some robots with solar!

Creative Process: Building Overhead Video Supports

Finished (but not painted) upright floor joist clamp and horizontal support bracket.

"I love it when a plan comes together."

I love the whole process, actually, from conceptualization to execution. I often fizzle before getting to execution, so it's particularly sweet when I get there. The creative process is no mystery. Processes are not generally a mystery, by definition.

Step 1: Have an idea, an urge, or a problem. Define it. I find it helpful to start with words.

Problem, I need a way to secure my expensive video camera over my work area while I'm shooting in the studio... I mean basement. 

Step 2: Ideation. This part has no tangible output. It's not something you can force, either. You just have to give yourself room to think about the problem. That usually means time. Don't try to think about it either, or at least don't keep trying. Do other things, especially menial things. Thoughts will come to you and you just need to let it happen. The danger is that you give yourself so much time that you stop thinking about it even subconsciously. It's hard to gauge, but eventually you get the hang of it.

Since I have Camera A shooting from the front, I don't want a tripod in the picture. Also want the camera looking straight down. First thought is to use bar clamps on the floor joists and suspend a rod or 2x2 between them, but that scared the dickens out of me. 

Step 2: Visualize the idea (thumbnail)

Step 3: Repeat ideation, then refine idea.

Enough to go on before shopping for materials. Bought stuff at Home Depot, and got to thinking. Why cut the 2x4 when it already came in a handy 92" length? It would be stronger and easier to handle if it was just one piece. Got to drawing again.

Step 4: Test your idea. In design and engineering it's called prototyping, but in all creative endeavors there are ways to mock up your idea and experiment with the things that bother you the most from your refined idea. What part would you like to validate?

Step 4: Begin working. Now you have to be a bit fearless. You need to be confident that as things progress you can solve unexpected problems. You have a plan, and that should give you courage. That all sounds touchy-feely, but it's the honest truth. Psychologically that's the state we're in (i.e., fear) at this stage. So, pshych yourself up and get at it.

I glued two pieces of 2x4 together to form the clamp, secured at the bottom with a bolt and tensioned with another bolt through the hole in the picture below. This way the upright support will come in just two pieces, and will be easier to handle, like I'd hoped.

Step 5: Solve problems and strive for simplicity.

In the drawings I had considered what to do with the two bolts going through the bracket. What I drew was a inset in the middle piece of wood. While working I realized this would be difficult to align, and I didn't need two bolts going through the clamp at the bottom. In fact, that might make the clamping less effective. So, first I cut the clamp short enough that it would be above the bolt. Then I realized that the bolts I'd gotten weren't long enough to go all the way through the 2x4 and the clamp anyway. I routed out the cavity that was going to be in the middle piece in the first place, but in the main board itself. Then I realized that with this design I could leave the clamp loosely joined to the main support and it can swivel down, making setup easier. Simple solutions tend to be more elegant.

Step 6: Finish. Don't stop. Don't skimp. Don't cut corners. Fight impatience. Don't be a perfectionist either. Get it done and let it be.

It worked! Set it up and called Desi down to see what I'd made. Just need to let the glue dry and paint it all white. Then I'll be done. The camera will be suspended from the horizontal 2x2 by a Joby Gorillapod SLR-Zoom. $35 on Amazon. 


Step 7: Enjoy your success. You did it!

Part Specifications Matter

1. Stepper motors arrived today. Yay!
2. Nuts and bolts. Now I think kits are the way to go because of this stuff.
3. Bag o' printed parts off eBay. They were the first to arrive.
4. Rod. Not as hard to get as I thought. Fastenal had the drill rod.
5. Spirit level, yard stick and calipers. Measure THEN cut. :)
6. Instructions. Won't actually use iPad around dust.
7. Springs. 12 to a box. Need less than one whole one.
8. More hardware, arrived today. Need a new box.
9. Accelerant (Mountain Dew)
10. WD40 -- eternally useful

Almost there! Still missing: electronics, hot end, pesky unspec'ed springs, and ... 5/16" washers, damnit!

True Value let me down. Not their fault entirely, I suppose. I asked for "5/16 in. washers" and they gave them to me. What I failed to note, however, was that the ones needed for the project have an outer diameter of 5/8" and the ones they handed me have an OD of 7/8". So, while try-fitting some of the parts tonight I realized that I'd gotten the wrong ones, and the ones I have won't work. So there's another $5 I didn't need to spend. Anyone need some washers?

Moral of the story: pay attention to all the exact specifications of the project, even when you think something is a no-brainer. 

Build Your Own CNC

Down time is a dangerous thing when you're hot to build a project. I am waiting for delivery of a whole bunch of parts for the RepRap machine project that I bought online Monday, and wanting to take a break from thinking of that particular project, got surfing.

My next project has been selected, I think. It was a toss-up from the beginning whether I would do the RepRap 3D printer or build a CNC mill first. The RepRap won because as it turns out, it's what I could get funded first. These kickstarter projects, Modular Desktop CNC Machine, and DIY Desktop CNC Machine have recently added fuel to the fire, though. They lack one important ingredient, however, and that's that neither as far as I can tell is open source. It's important to me that I build from a base that I can share after it's done. These machines I'm building are going out into the community, and if people are inspired by the projects I want to be able to share time, experience and materials with them. I could probably do that with the DIY Desktop CNC Machine project by Stephen McGloughlin, but I know that the other project is NOT open source at all because I asked and they said "no".

Come to think of it, though, there are actually two things. I also envisioned a CNC mill that was as large as Ben's on the Ben Heck Show. The Kickstarter projects are just tabletop machines, and don't scale up. I probably will end up building a desktop CNC mill at some point. There's no sense having a 4' x 8' cutting table when you're working on a 2" x 3" PCB after all, but I'm definitely looking at one large enough to cut cabinetry and other larger-scale pieces.

So, looking back on the Kickstarter projects I was actually considering their respective merits, supposing I would go with one of them anyway, but gave "open source diy cnc" a whirl on the Google machine. After trying quite a few, and lingering a while at diylilcnc.org I eventually stumbled on the motherload, buildyourowncnc.com. The reason I found it so exciting right away is that this project starts with, as the author puts it, "a cheap saw, cheap drill, cheap metal cutting saw, a 5/16" tap (a device that threads the inside of a hole to enable a screw/bolt to fasten) and a cheap screwdriver. When I say cheap, I mean it literally, and economically."

Plus, the guy that developed this project is just cool.





In the videos on his web site where he's showing all the steps of the process you hear his two kids crying or making noises while they play in the background, and his wife giving him a hard time. At one point you hear her say something like "...it's been a whole year..." in a very exasperated voice and he has to remind her, "You're being recorded, honey." You can imagine she's been complaining about the project because the whole thing is being built on the guy's kitchen table in what might be essentially a one-room apartment by the looks of it. The toxic MDF routing is done in his bathroom to protect the family and make clean-up easier. This guy was seriously dedicated, or driven, or whatever.

He wrote a book about the project, a DIY manual for repeating his effort and building your own bootstrap CNC mill. I just bought the book and can't wait to read it. Link below.