A few weeks ago a friend highlighted an online course Control of Mobile Robots through Coursera for me to check out. I was intrigued because of the content (obviously) but also because I’d not yet experienced one of these “online universities” such as Khan Acadamey or Uda City. In fact I’d never even heard of Coursera before.

I just finished my first week of class. The sessions were good, the video quality was outstanding, and the material was interesting. The format of this particular course is set up to be a series of short videos (6-10 minutes each) plus a teaching assistant video (called a “glue lecture”) that is intended to help with the weekly quiz. There is also a forum for information exchange and Q & A.

I really enjoyed the lectures, very interesting material and the content is easy to follow. I did find the “glue lecture” was very different than the regular lectures, in as much as it was very focused on math especially differential equations. I’m really glad that I also checked out the forum before starting the quiz. There was a little bit of feedback like “wow this requires more math than expected” and “where to get enough background in differential equations to understand this course?” so I knew I needed to brush up. Fortunately Google is full of useful information. Another hour or so doing some background reading and playing with some graphing tools (like Wolfram Alpha and Excel) I was feeling more confident.

After re-watching the “glue lecture” a few times, and working out some of the examples on paper, I decided to take the quiz. I was well prepared because it was definitely focused on the math. The questions were good, easy to understand, and (for a multiple-choice test) it was well designed to insure guessing was not going to get a passing grade.

Week one is now “in the books” and I got 100% on the quiz. Yay me!

I’m impressed with Coursera so far, and would recommend it if interested in giving yourself a bit of a mental challenge with a structured educational experience.

These pictures show the final result of my 3D printing experience. Its a bracket to hold four Parallax Ping))) ultrasonic sensors. This replaces the original three-sensor mount that I fabricated myself from Delrin and Sintra parts, using my Sherline mill. You can see that in the videos in this post.

I created the model in Blender. I can’t really say too much about Blender, other than its best attribute is that it works for what I needed it for, the price is great (free), and its a myth that Blender can’t create parts to specific sizes (it can, it just doesn’t do it by default, and the correct setting isn’t obvious). Its complicated to use, but it did work for me (eventually). It probably took me about 20 hours to learn the software, experiment with it enough to design my part, and create the final model as you see it. There are many, many tutorials online and I’d recommend starting with those if you want to learn anything about Blender.

I printed my parts at Shapeways, a commercial 3D printing service. The service was excellent, very similar to using BatchPCB e.g. create an account, upload the design, pay by credit card, then wait for the results in the mail. My rationale for using Shapeways instead of something like the Makerbot is economics and quality. Although quick turnaround time would be great, the cost plus hassle plus poor quality of the DIY printers (like Makerbot) just isn’t worthwhile to me. I could print 20 of these brackets for the price of one Makerbot. That doesn’t include the cost of my time and frustration figuring out the “science project” required for great results when operating these machines. Its the same argument for why I use BatchPCB instead of making my own PCBs.

The pictured item is actually my second attempt at a 3D print. My first attempt was successful (in as much as it was printed exactly as it was designed) but it wasn’t really robust enough. I learned a lot about wall thickness, joins, etc. In addition to making it stronger, I also improved the design such that no extra pieces were required. If you look closely at the final part, you’ll see that the sensor board mounts are designed right into the part. Just tap the hole then use 4-40 screws to attach the Ping))). So much more convenient than the original part I fabricated myself that required multiple brackets.

My summary: 3D printing is a fantastic technology for making certain types of parts, especially complex shapes from plastic. After a lot of research and observations of people using the Makerbot (and similar machines) I concluded that these DIY printers are too finicky. I’m sure the technology will improve, and I believe I will eventually buy/build a “home shop” 3D printer, but for today Shapeways is a much better deal.

By the way, Shapeways offers metal printing as well as more traditional plastics, but the price is quite high, and I don’t have high hopes for it being as sturdy as something I’d make by milling down a block of metal myself. But 3D printing makes certain designs a lot easier to create. My milling machine and lathe will still get used for many future projects, but sometimes certain objects lend themselves to 3D printing a lot better.

Went to the SRS Robothon 2012 this weekend in Seattle. Same as in the past few years I helped organize the event, although most of effort is just doing stuff with the website and online activities; participating from Vancouver is otherwise a bit challenging.

I ran the Robo-Magellan contest this year, and I had a great time. I wasn’t able to finish a robot for the contest myself, but several others made a valiant attempt. The entries were a bit less than stellar, but its encouraging to see people continuing to build for this challenging event. I want to figure out a way to inspire more people to enter, as it is a big leap ahead in complexity and should therefore be a logical “upgrade” from other events.

I did enter two robots in the Minisumo event, and managed to place 1st overall with The Thin Man. This robot is definitely a bit old now but still remains reasonably competitive. The second place finisher was a robot named Ram that is very competitive and even older than mine! My take-away from this was that: (a) not a lot of people are building new minisumo robots; and (b) state of the art hasn’t progressed much beyond what was good a few years ago. Bill Harrison made an appearance with several minisumo robots of his own, and although he didn’t place in the top three it was great to see the “father of minisumo” involved again.

The other two events Line Following and Line Maze got decent turnout as well, and as usual the crowd was great. The renovated facilities as Seattle Center are a big improvement, and the “tech fair” atmosphere that weekend was great to see.

Everyone seemed to have a great time, and the current committee was enthusiastic about another event next year. Right now it seems likely that there will be a Robothon 2013 next June! Hope to see you there.

Just booked flights to go to this year’s Sparkfun Autonomous Vehicle Competition in June. Should be a blast!

My Outdoor Robot project got some attention recently, now utilizing the ultrasonic sensors at the front to avoid obstacles. I’m still running it very slowly, so I can learn from watching it without having to run to catch it. I’ve implemented a subsumption architecture in the software that has a few simple behaviors:

1. Startup: waits for a button press, and only allows for one minute runtime.
2. Avoidance: using the front-facing ultrasonic sensors, turn away from obstacles. The turn is proportional to the distance to the obstacle.
3. Cruising: run straight ahead at the maximum (slow, at this point) speed.

In the videos below you can see this happening. There really isn’t a lot of “smarts” to the software, although it looks pretty good while running. Next up is to get the “remote stop” working then I’ll increase the speed. I also need to work on behaviors for when the robot gets itself into a corner e.g. backup and turn around.

Enjoy these videos for now!